Source: https://oguzerol.com/en/autonomous-ai-systems-glossary/

# Autonomous AI Systems Glossary

[Oğuz Erol](https://oguzerol.com/en/about-me/)  |  98 min read  |  Published: 21 July 2026

## Quick summary

A guide to AI agents and autonomous systems. Learn the core concepts and technical terms used in the world of agentic AI, with clear explanations.

## .antigravityignore

This is the physical blacklist file that stops the agent from getting lost in giant libraries like \`node_modules\` while working across the file system and suffering a [Segmentation Fault (memory overflow)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#segmentation-fault-bellek-tasmasi).

It works on the same logic as a traditional \`.gitignore\` file; but where \`.gitignore\` only [ignores](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme) changes on the [git](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git) side, \`.antigravityignore\` directly protects the agent's working area and the [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar). It ensures agents in [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) ([Antigravity IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-ide), etc.) environments only move around within the [Workspace Rules](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workspace-rules-yerel-kurallar) you've allowed, which both boosts speed and prevents wasted [tokens](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b).

## 60/30/10 Rule

This is an [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) principle used to optimize resource and budget management across agent ecosystems. Instead of processing every task in the system with equal weight, it splits work into 3 main categories:
- 10% (Complex): Top-tier [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) are used for tough tasks that need architectural decisions and deep [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme).
- 30% (Medium): Standard [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) are used for routine work like data processing or drafting.
- 60% (Simple): Low-cost [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) ([Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics)) are preferred for simple jobs like code formatting or [routing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#redirect-yonlendirme).
This rule balances intelligence against [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi).

## A2A (Agent-to-Agent)

This is the communication protocol where agents talk to each other directly (P2P) without waiting for human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) approval or going through a central [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan). Unlike centralized structures, it lets every agent in a [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) share data autonomously with the others.

For example, when an [Implementer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementer-agent-uygulayici) hits an error, instead of reporting the problem to the [main conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef), it contacts the [Resolver Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#resolver-agent-cozucu-ajan) directly. During this exchange, [Identity Propagation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#identity-propagation-kimlik-akt) keeps permissions intact end to end.

## ACL (Access Control List)

This is the hardware-level security barrier that [restricts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) agents and sub-components like [Sidecars](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sidecars-yan-surecler) from freely [reaching](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) the operating system's files, folders, and network resources. Agent security is handled in 3 layers:
- At the high level, [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) (prompt [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar))
- At the communication level, [MCP](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) [restrictions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis)
- At the hardware and system level, the ACL.
ACL rules designed on the [Principle of Least Privilege](https://oguzerol.com/en/autonomous-ai-systems-glossary/#principle-of-least-privilege) make sure that even if a malicious [Prompt Injection](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-injection-zehirli-girdi) attack succeeds, the agent is still flatly blocked from deleting critical system files.

## ACT

This is the **"apply and trigger"** stage where autonomous AI doesn't just think, but makes physical or digital changes out in the real world.

It's one of the 4 core phases of the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) cycle, turning the decisions made during [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) and [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) into concrete action.

At the moment of action, the agent uses [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) — its hands — to write a record to a database, create a file, or trigger an [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api).

## Adversary Agent

This is a cybersecurity ("[Red Teaming](https://oguzerol.com/en/autonomous-ai-systems-glossary/#red-teaming)") agent tasked with attacking your own system in order to test the resilience of autonomous systems. Unlike constructive (generative) agents such as the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) or a [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan), it plays a simulated attacker role focused on finding the system's [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) and security holes.

Its job is to push against [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) mechanisms, try to force agents past their [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) with [Prompt Injection](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-injection-zehirli-girdi) attacks, and report every vulnerability before the system goes live.

## Agent OS

This is the intelligent [orchestration](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestration-orkestrasyon) layer that manages the memory, hardware resources, and task queues of hundreds of autonomous agents in the [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) — much like a Windows or Linux kernel.

It sits above [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) structures, which focus on the individual operations of a single agent. It controls agents' [Principle of Least Privilege](https://oguzerol.com/en/autonomous-ai-systems-glossary/#principle-of-least-privilege) rights, kicks off [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) (automatic recovery) when something crashes, and provides a secure [hosting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hosting-barindirma) service for [A2A](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent) communication between agents.

## Agent Routing

This is the process of analyzing a complex request coming from a user or the system and autonomously routing it to the right specialist [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) unit. It reads the intent of the request using [Semantic Routing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-routing-anlamsal-yonl) instead of traditional if/else blocks.

It's run by the central [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan). For example, a request that needs a database query gets sent straight to the [SQL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sql) specialist agent, while a design change is assigned to the [CSS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) specialist agent.

## Agent Skills

These are expertise (know-how) packages that go beyond just giving the agent a list of [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri), and instead explain "how and in what order" to actually use that tool in the real world. They're usually documented in [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd) format.

They deliver not just code but an institutional way of working. By teaching agents not only what to do ([System Prompt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#system-prompt-sistem-istemi)) but how to do it ([Procedural Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#procedural-memory-prosedurel-hafiza)), they guarantee deterministic, predictable results.

## Agentic AI

These are autonomous AI systems that take just the end goal from you, build their own plan behind the scenes, correct themselves when they make a mistake ([Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma)), and act using tools.

Unlike passive chatbots such as [ChatGPT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chatgpt), it analyzes the problem with [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme), builds a [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla), and actually does and completes the work within the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) cycle.

Thanks to its [Orchestration](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestration-orkestrasyon) ability, it can coordinate multiple specialist [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units.

## Agentic Engineering

This is the discipline of building autonomous architectures — with mathematical [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) designs, rigorous [Evals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#evals-evaluation---hard-evals) (tests), and impenetrable [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) firewalls — in place of fragile, luck-based, primitive coding practices ([Vibe Coding](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vibe-coding)).

It's the art of building production-ready, enterprise-grade multi-agent systems that don't crash in the face of errors, but instead push forward by resolving their own mistakes through [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma).

## Agentic Loop

This is the endless spiral life cycle at the heart of an autonomous system, [iterating](https://oguzerol.com/en/autonomous-ai-systems-glossary/#iterate-yinele) on itself until it reaches the goal. It makes the agent follow these 4 phases at every step:
- [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet)/[Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle): [Observing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) and analyzing the environment.
- [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla): [Planning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) the steps to take.
- Reason: Internal reasoning and decision-making.
- [Execute](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula)/[Act](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem): Taking [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) in the outside world.
Unlike classic code flows, this cycle is dynamic — when it hits an error it loops back to the start and develops new strategies using [Loop Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#loop-engineering-dongu-muh) principles.

## Agentic RAG

This is an autonomous data-retrieval loop that works actively, like a detective, instead of the static [RAG (Retrieval-Augmented Generation)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) approach that blindly accepts whatever's pulled from a [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani).

The agent tests the accuracy of the data it finds with [State Verification](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-verification-durum-dogrulama), and if it finds something missing, it kicks off a new search with different keywords on its own. The goal is to raise [Grounding](https://oguzerol.com/en/autonomous-ai-systems-glossary/#grounding-kaynaga-baglama) quality and drive [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) risk down to zero.

## Agentic IDE

This is an integrated development environment where agents work shoulder to shoulder with developers, given full powers like writing code, [Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim), reading files, and [MCP Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-client-mcp-istemcisi) integration (e.g., Cursor, [Antigravity IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-ide)).

Unlike browser-based chatbots, they run [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) and [Execute](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) cycles autonomously, directly inside your project, on your local files (within the allowed [ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar)).

## AGI (Artificial General Intelligence)

This is the hypothetical level of super-intelligence that goes beyond today's [Narrow AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#narrow-ai-dar-yapay-zeka) systems — which specialize in specific domains like writing code or translation — and can teach itself to do any cognitive, creative, or analytical task a human could do.

Agent systems, which combine different [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) specialties under a single [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan), are regarded as some of the most important building blocks on the road to AGI.

## Always Proceed

This is an aggressive permission setting that lets the agent keep going autonomously — without asking a human for [HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop) approval — when it's about to run commands that could be risky or change the system (e.g., deleting a database, running [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) code).

While it speeds things up inside Autonomous Mode, when it isn't backed by [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) it can create serious risks to system security.

## Antigravity

This is an advanced agent command center that lets hundreds of AI agents be coordinated like a massive enterprise software team.

By housing [Antigravity CLI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-cli-agy) (the [Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) tool), [Antigravity SDK](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-sdk) (the software library), and [Antigravity IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-ide) (the development environment), it lets developers manage and supervise autonomous processes end to end and build a fully integrated [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru).

## Antigravity 2.0

This is the next-generation [desktop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) version of the [Antigravity](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity) platform, with a graphical interface (GUI) that goes beyond the classic [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) ([CLI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-komut-satiri-arayuzu)).

It visualizes the complex output agents produce in interactive [Artifacts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar) panels, lets you watch the [Subagent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan) hierarchy live on a graph, and lets you visually step into the [Chain of Thought](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chain-of-thought-cot) (thinking stages). It's a professional control center where the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) and its [sub-agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan) are managed with the ease of [drag](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru)-and-drop.

## Antigravity CLI (agy)

This is a lightweight, text-based agent controller with no graphical interface (GUI), built to run directly inside SSH [sessions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) or IDE [terminals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) for developers.

It consumes almost zero RAM and processor resources. Working on [TUI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tui-terminal-user-interface) ([Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) UI) logic, it's a fast, minimalist tool that lets you assign autonomous tasks to background agents, list their status, and manage system [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) processes.

## Antigravity SDK

This is the core architectural library that lets developers build their own custom autonomous agents in Python or TypeScript, add new [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) to the system, and write OS-level [JSON Hooks](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-hooks-json-kancalari) triggers.

Where [Antigravity CLI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-cli-agy) manages things externally through commands, the SDK gets embedded directly into your [application's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) codebase, letting you build your own [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) infrastructure.

## Antigravity IDE

This is a modern development environment purpose-built for writing code, reviewing, and running processes together with agents, combining [Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal), [MCP Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-client-mcp-istemcisi), and autonomous system components under one roof.

Unlike an ordinary editor, while you're writing code, it lets agents with [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) abilities access your file system and work with you in an asynchronous, [A2A](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent)-style sync (peer-programming).

## API

This is the traditional data interface where different pieces of software and servers talk to each other with fixed rules (endpoint, request, response).

Unlike fixed, static classic API connections, the dynamic and flexible [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) lets AI autonomously [discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) tools and resources.

In older systems, the agent needs an [API Connector](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api-connector) bridge to be able to talk to an API at all.

## API Connector

These are classic translator (bridge) plugins that let an agent talk to the outside world on legacy closed systems (e.g., Salesforce, legacy ERPs) that don't directly support the modern [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) architecture.

They give agents flexibility by translating the complex XML or [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) data an [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) returns into a simple [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) capability the agent can easily understand and process (e.g., a \`fetch_user_data\` function).

## Artifacts

These are the outputs an agent produces after long, complex operations, presented not as boring plain text but in a human-readable, visual (structured) format — an [Implementation Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementation-plan), a UML diagram, a table, or a Markdown report.

This approach makes human [observation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) easier and lets the developer quickly approve ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) the system's decisions and keep the process moving.

## BigQuery

This is a massive data-analysis engine running on the [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) Cloud platform that can query billions of rows of data in seconds.

Autonomous agents can export Google Analytics (GA4) data into BigQuery and run deep analysis with [SQL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sql) queries. In [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) workflows, it's a powerful [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) resource that turns raw data into meaningful insight.

## Browser Sub-Agent (/browser)

This is a visual [sub-agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan) ([Subagent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan)) coded specifically for web interface testing. It connects to a browser autonomously through Chrome DevTools and clicks buttons and fills out forms like a human.

While running [UI Testing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ui-testing-arayuz-testi), it autonomously records video (WebM) to prove any errors it hits, and reports the status back to the top-level [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) along with the evidence.

## Budget Guardrails

These are hardware-level safety fuses that stop an agent from falling into an infinite loop because of an unexpected error and blowing up costs in seconds with a flood of [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) requests.

They include [Cost Control Limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cost-control-limits). Standard [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) are usually about security and [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) (read/write), while Budget Guardrails are entirely about policing the agent's [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) [limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) and hard-stopping the system in an emergency when needed.

## Campaign Syncing

This is the process of agents autonomously syncing marketing campaigns (Google Ads, Meta Ads, etc.) with external platforms using an [API Connector](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api-connector).

It replaces manual copy-paste work. The agent updates the ad copy out on the external platform during the [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage, based on rules or budget strategies that change in the database (e.g., [Firestore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#firestore-firebase)).

## CDN (Content Delivery Network)

This is speed infrastructure that distributes a website's static files (images, [CSS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript), [JavaScript](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript)) across servers worldwide, serving each user from the closest point (e.g., Cloudflare, AWS CloudFront).

It directly affects [Core Web Vitals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#core-web-vitals) and [PageSpeed / Lighthouse](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pagespeed---lighthouse) scores. Autonomous agents can suggest CDN configuration changes based on the results of speed tests.

## Centralized Agent Networks

This is the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) in a multi-agent architecture ([MAS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru)) where the network is managed strictly and hierarchically by a single central "dictator" agent in the [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef) role.

It's a structure where lower-level [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units don't make autonomous decisions on their own, only [carrying out](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) the [Task Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#task-plan-taskmd) steps they're ordered to. It runs slower than the [Decentralized Agent Networks](https://oguzerol.com/en/autonomous-ai-systems-glossary/#decentralized-agent-networks) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), but it's safer for enterprise [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim) and auditing.

## Chain of Thought (CoT)

This is when an [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin), instead of jumping straight to a conclusion and making a mistake ([Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon)) while solving a complex problem, "thinks out loud" — in speech or text — by breaking the problem into steps (e.g., "Step 1: find x, Step 2: multiply by y").

It's the core strength of advanced [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model). It lets the agent convince itself and reach error-free conclusions by following a logical order during the [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) stage.

## Chat Room Debate

These are multi-agent autonomous decision-making sessions where agents form [Debate Loops](https://oguzerol.com/en/autonomous-ai-systems-glossary/#debate-loops-tartisma-odalari) and clash ideas from different perspectives, much like a WhatsApp group.

Typically an Implementer, a Reviewer, and a [Contrarian Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#contrarian-agent-aykiri-ses) are put in the same room ([Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi)). This keeps the system from blindly accepting the first solution it finds, and instead lets it reach the right decision through consensus.

## Chatbot

This is an automated chat assistant on websites, usually rule-based (if/else) or backed by simple AI, that answers visitors' questions.

The difference from [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems is that it can't step outside predefined scenarios, can't use [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri), and can't make autonomous decisions. Modern agentic systems are equipped with [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) and [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) abilities that go far beyond a chatbot.

## ChatGPT

This is the AI assistant built by [OpenAI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai) that millions of people use directly through conversation. Its core difference from [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems is that it can't run an autonomous cycle ([Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu)) and can't physically intervene in the outside world with [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri).

ChatGPT is a passive chatbot; but in agentic systems like [Antigravity IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-ide), that same [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) brain turns into an active machine that runs autonomously through [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) and [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) cycles.

## ChatSessionStore.index

This is the local [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) index database that links an agent's old chat [sessions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) to the project's [State Management](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-management-durum-yonetimi) history, especially in [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) ([Antigravity](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity), etc.) systems.

It's not a temporary [Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek) file; it's a sensitive, critical-to-back-up file that, when it gets corrupted or throws a read error, makes an agent's past ([Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek)) experience unreachable ([Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi)).

## Claude (Anthropic)

This is the AI [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) family (Claude Sonnet, Opus, etc.) built by Anthropic, focused on safety and harmlessness. It's a direct rival to [OpenAI's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai) GPT series.

Thanks to its long [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) capacity (200K+ [tokens](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b)), it can analyze massive codebases in a single pass. In [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) ecosystems, it's used as an alternative [brain](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) for [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) or [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) tasks.

## Clean Architecture

This is the standard where software is designed in layers ([Domain](https://oguzerol.com/en/autonomous-ai-systems-glossary/#domain-alan-adi), Use Cases, Interfaces), fully isolated from external dependencies (UI, Database, Framework).

When agents write code, they check Clean Architecture rules against [Semantic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-memory-anlamsal-bellek) stored in [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd) documents to prevent [Technical Debt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#technical-debt-teknik-borc) from building up. Enterprise agents take these rules as a template so they don't produce spaghetti code.

## Client (MCP)

In the new-generation [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) architecture, this is the bridge (communication) layer that runs directly inside the application ([Host](https://oguzerol.com/en/autonomous-ai-systems-glossary/#host-ev-sahibi---mcp)) and lets the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) talk to the [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) on the other end.

It autonomously discovers the tools the server offers through Capability Negotiation, and plays the role of go-between by translating the [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) simple [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) requests into the standard [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)-RPC commands the server understands.

## Cloud Functions

These are small pieces of code that run automatically when a specific event fires (a Trigger), without needing any server management (e.g., AWS Lambda, [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) Cloud Functions).

They run background tasks triggered by autonomous agents via a Webhook or [JSON Hooks](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-hooks-json-kancalari) cheaply and at scale. They're the building block of [Serverless](https://oguzerol.com/en/autonomous-ai-systems-glossary/#serverless-sunucusuz) architecture.

## CLI (Command Line Interface)

Without a graphical interface (GUI), this is the fast communication layer where agents manage a server, a local machine, or an [application](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) directly with text-based commands (bash/PowerShell).

It's the most natural interface for agents. Compared to [Vision](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme)-based interfaces that have to interpret pixels, it consumes far fewer [Tokens](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b) and improves the system's [Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti) quality by getting results back deterministically (\`--[json](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)\`).

## CLI Tools

This is the group of low-level tools ([Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri)) that let an agent intervene directly at the operating system's [Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) command line ([CLI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-komut-satiri-arayuzu)).

For example, it covers capabilities like \`npm install\`, \`[git](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git) commit\`, or \`mkdir\`. Because it grants direct write [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) to the system, using it without isolation behind [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail) ([Sandbox](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox)) or a strong [ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) barrier is very risky from a security standpoint.

## CMS (Content Management System)

This is a platform that lets you manage a website's content (posts, pages, media) without needing technical knowledge. WordPress, Webflow, and Wix are the most common examples.

Autonomous agents can connect to CMS systems through an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) to autonomously handle content creation, SEO optimization, and page updates.

## Compound AI System

This is a "compound" architectural approach made up of a chain of multiple small/specialist [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), an external [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani), and dynamic [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri), instead of loading all the work onto a single giant (and expensive) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) the way traditional systems do.

The agent itself isn't a single [brain](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin); instead, it solves complex problems as an autonomous whole by using a Compound AI System ([RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) + [Model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) + Python Engine).

## Conductor

This is the strategic manager sitting at the very top of the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan)-Worker hierarchy. It doesn't waste time writing code directly ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) or connecting to databases.

Its job is to break a big incoming request (e.g., "Refresh the site") into pieces ([Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)), distribute tasks to the [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units below it, and combine their sub-outputs to approve the final product. In the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), it takes on the "Checker" role.

## Context Engineering

This is the art of going beyond just using magic words for [Prompt Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-engineering) to boost system quality, and instead designing exactly which specific, clean, up-to-date [RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) data gets placed in front of an agent at runtime.

The focus is on trimming away unnecessary data (Context Pruning) to protect the agent's [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) limits and prevent [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) syndrome.

## Context Rot

This is when an agent forgets the actual topic (task) at hand because its memory space ([Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi)) has piled up with massive, irrelevant code files and repetitive, long error logs ([Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle)).

As a result of this memory poisoning, the agent becomes paralyzed, its response time slows, and it slides into [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) (making things up), crashing the system. The fix is to regularly summarize the history and clear out the [Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek) (short-term memory) space.

## Context Wall

This is a data-filtering system set up in large software projects to keep lower-level [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units from getting crushed under the entire project's massive data ([Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi)) and making mistakes.

The agent is given an isolated (walled-off) cell where it's only shown "just enough information to do its own specific task." This way the agent doesn't drown in unnecessary libraries and doesn't run into Focus problems.

## Context Window

This is the maximum **"short-term working desk"** capacity an AI model can lay out in front of itself and process at once.

Once that desk fills up, the AI starts forgetting older information; in long conversations this creates the risk of [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi), and over time, [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi).

To get around this memory limit, architectures use [RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) or [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek).

## Contrarian Agent

This is a critical agent profile inside [Debate Loops](https://oguzerol.com/en/autonomous-ai-systems-glossary/#debate-loops-tartisma-odalari) that deliberately opposes and objects to the ideas and code other agents produce.

It stress-tests from the inside with [Red Teaming](https://oguzerol.com/en/autonomous-ai-systems-glossary/#red-teaming) logic. Its job is to find weak spots (edge cases) to keep the team from getting complacent and push it toward perfection. Where a standard [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) looks for compliance, the Contrarian Agent looks for weakness.

## Copilot (GitHub/Microsoft)

This is the AI assistant built by Microsoft and [GitHub](https://oguzerol.com/en/autonomous-ai-systems-glossary/#github) that gives developers instant code suggestions and chat support.

Its core difference from [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) systems (e.g., Cursor, [Antigravity IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-ide)) and their traditional [versions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) is that it's focused on code completion, can't start its own autonomous cycle ([Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu)), and can't act independently in the outside world like a [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan). Even though newer plugins push its [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) further, at its core it's still an assistant [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) tightly bound to human commands.

## Copywriting

This is the art of writing sales-focused, persuasive text with a call to [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) (CTA). Landing page headlines, ad copy, and email subject lines all fall under this discipline.

Autonomous agents can analyze A/B testing data to see which tone of copy drives higher Conversion Tracking and generate content recommendations from it.

## Core Web Vitals

These are [Google's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) 3 core metrics for measuring a web page's user-experience quality:
- LCP (Largest Contentful Paint): How long it takes the page's main content to load.
- INP (Interaction to Next Paint): How fast the page responds to a user's click.
- CLS (Cumulative Layout Shift): How much the page's visual elements shift around.

They directly affect SEO rankings. They're measured with [PageSpeed / Lighthouse](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pagespeed---lighthouse) [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri). Autonomous agents can analyze these metrics and suggest [CDN](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cdn-icerik-dagitim-agi) and code optimizations.

## Cost Control Limits

These are hourly/daily financial barriers and caps placed on autonomous agent platforms.

They stop an agent that's fallen into an infinite loop due to an error from firing off hundreds of unnecessary [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) requests ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) back to back and blowing up its [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi). They form the numeric rules of the [Budget Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#budget-guardrails) architecture; when a limit is exceeded, the system automatically cuts off the agent's [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) to prevent a blowout.

## Crawling

This is the process where [Google's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) and other search engines' bots ([Googlebot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google)) systematically move through websites, read their content, and save it to a database ([Indexing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#indexing-indeksleme)).

The [robots.txt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#robotstxt) file determines which pages get crawled, while the [Sitemap](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sitemap-site-haritasi) lists which pages should be crawled. Autonomous agents can spot and fix crawl errors through Google Search Console.

## Critic / Reviewer Agent

This is an independent quality-control agent built purely for auditing and review, taking on the "Checker" role in the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model).

When the [Implementer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementer-agent-uygulayici) finishes its work, this agent looks at the output with a blank, unbiased memory and checks whether it meets the standards. If it approves, the code gets merged; if it finds an error, it sends autonomous feedback for a fix.

## Cron (Scheduled Tasks)

This is scheduling infrastructure that lets agents wake themselves up and run tasks entirely in the [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) ([Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis)) at fixed intervals (e.g., every hour), without human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) approval.

For example, an agent scanning the database every night to generate a report, or running [Campaign Syncing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#campaign-syncing) every hour, is handled by cron triggers at the [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) level.

## CSS / HTML / JavaScript

These are the web's 3 core building-block technologies:
- HTML: Builds the page's [skeleton](https://oguzerol.com/en/autonomous-ai-systems-glossary/#harness-kosum-iskelet) (structure).
- CSS: Determines the page's appearance (design, colors).
- JavaScript: Adds interactivity and dynamic functionality to the page.

Autonomous agents use these 3 languages inside an [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) to produce websites with [Responsive Design](https://oguzerol.com/en/autonomous-ai-systems-glossary/#responsive-design-duyarli-tasarim). A [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) then autonomously tests this code through the [DOM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dom-document-object-model).

## CSV / JSON

These are two common formats used for moving and storing data:
- CSV (Comma-Separated Values): A simple text format that stores table data separated by commas. Campaign reports, product lists.
- JSON ([JavaScript](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) Object Notation): A format that stores [structured data](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-markup-yapisal-veri) as key-value pairs. The base language of [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api), [MCP](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol), [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling), and [Schema Markup](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-markup-yapisal-veri) communication.

## Curate / Curation

This is the process of taking the thousands of pages of complex data agents find (through [RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) and similar methods) and, instead of sending it straight to the main [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), weeding out the junk and shaping it into something synthesized and contextually appropriate.

By filtering out unnecessary data (Context Pruning), it keeps the agent from suffering [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) syndrome. Uncurated raw data wears down the agent's memory and degrades its quality.

## CI/CD (Continuous Integration/Deployment)

This is the automated pipeline for a piece of software, running from being coded to being tested and pushed live ([Deploy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama)) — Continuous Integration / Continuous [Deployment](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama).

Within the [SDLC (Software Development Life Cycle)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sdlc-yazilim-yasam-dongusu), autonomous agents can have a [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) audit the code, and once it passes tests, autonomously [deploy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama) it through the CI/CD pipeline. It works integrated with [GitHub](https://oguzerol.com/en/autonomous-ai-systems-glossary/#copilot-github-microsoft).

## Daemon

These are ghost agentic ([Sidecar](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sidecars-yan-surecler)) processes that listen 24/7 in the operating system's [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) and wait for commands, even after the [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) window is closed.

They stay awake continuously and watch for events in the system. For example, triggering systems that instantly detect when a new file lands in the system and autonomously spin up a [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) in response depend entirely on Daemon processes.

## Dashboard

This is a monitoring screen that summarizes [KPI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#kpi-anahtar-performans-gostergesi) metrics, campaign performance, and site traffic with visual charts, tables, and maps.

It's built in natively on platforms like Google Analytics (GA4), Google Ads, and Meta Ads. Autonomous agents can build custom dashboard reports in [Artifacts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar) format to give humans ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) a quick status summary.

## Data Layer

This is the on-page [JavaScript](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) data layer that lets a website communicate with Google Tag Manager (GTM). User [actions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem), product info, and page data all get written to this layer.

It's the foundation for Tag and Trigger configuration. Autonomous agents can detect missing Data Layer variables and suggest fix code.

## Data Privacy

This is the legal and ethical protection principle [applied](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) to how users' personal data (name, email, location, behavior) is collected, processed, and stored.

It's regulated by laws like [GDPR](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gdpr-kisisel-veri-koruma) (Europe) and [KVKK](https://oguzerol.com/en/autonomous-ai-systems-glossary/#kvkk) (Turkey). Cookie Consent and Consent Mode are the technical [implementations](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) of this principle. Autonomous agents protect privacy in [Multi-Tenant](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-tenant-coklu-kiraci-izolasyon) systems by enforcing data isolation.

## Debate Loops

These are asynchronous processes where multiple agents argue with each other in a loop until they reach consensus on a shared design or piece of code.

These rooms bring together Implementer, Reviewer, and Contrarian agents, keeping the system from being stuck with the decisions of a single [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) ([Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon)) and autonomously raising output quality through multiple perspectives.

## Decentralized Agent Networks

This is a decentralized working [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) in multi-agent architecture ([MAS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru)) based on a flat (P2P) structure, where the network is managed without a central "dictator" agent like the [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef).

Agents don't wait for orders from outside; they make their own decisions autonomously and communicate directly with each other via [A2A (Agent-to-Agent)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent). It's more innovative than the [Centralized Agent Networks](https://oguzerol.com/en/autonomous-ai-systems-glossary/#centralized-agent-networks) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), but it needs stricter [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) against the risk of infinite loops or chaos.

## Deep Reasoning

This is when an AI, before producing direct output (an answer), spends hours or minutes in the [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) in a hidden loop, analyzing the problem by questioning itself ([Chain of Thought](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chain-of-thought-cot)) and building probability trees ([Tree of Thoughts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tree-of-thoughts-tot)).

It's the foundation of next-generation advanced [reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model). It's the exact opposite of the simple, speed-focused [Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics) approach; it's what solves the toughest 10% of tasks in complex [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) architectures.

## DeepSeek

This is a Chinese-origin AI [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) family that's open-source and low-cost, yet still delivers strong [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) capacity. Because developers can host it on their own servers (self-[hosted](https://oguzerol.com/en/autonomous-ai-systems-glossary/#host-ev-sahibi---mcp)), it offers advantages in [data privacy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#data-privacy-veri-gizliligi) and control over [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi).

It offers a strong, budget-friendly alternative to expensive [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin), and is often chosen as the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) brain in [Compound AI System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#compound-ai-system) architectures.

## Definition of Done

These are strict technical success (control) criteria that keep agents from leaving work half-finished or getting stuck in an unnecessary infinite loop after hitting an error.

They include fixed requirements like "tests must pass, code must be committed, and a [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) report must be returned." They let the agent know autonomously when to stop working on a task; the system doesn't leave the [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage until these criteria are met.

## Deliberative / Planning Agents

This is a type of methodical, thinking agent that, when it hits a problem, pulls back and analyzes the scope of the project first — instead of diving straight into code and taking [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) the way [Reactive Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reactive-agents-reaktif-ajanlar) do.

Before it moves, it prepares a blueprint ([Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)/architecture), presents that [plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) as an [Implementation Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementation-plan), and usually waits for approval from a human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) or a senior conductor before taking [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)).

## Deploy

This is the process of moving developed code or a website from the test environment to the live server (Production).

It's automated through [CI/CD (Continuous Integration/Deployment)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ci-cd-surekli-entegrasyon-dagitim) pipelines. Autonomous agents can have a [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) audit the code, commit it through [Git](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git), and trigger an automatic deploy. It's the final step of the [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi) criteria.

## Discover

This is the first step an agent takes when starting the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) process. It means [scanning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#crawling-tarama) folders and reading [environment variables](https://oguzerol.com/en/autonomous-ai-systems-glossary/#environment-variables-ortam-degiskenleri) to understand the task at hand.

During this prep stage, the agent uses [Dynamic Tool Discovery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-tool-discovery) to identify which [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) it has on hand, and builds context by doing a strategic situation assessment (discovery) before moving into the [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) stage.

## Docker

This is the technology that wraps [applications](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) and all their dependencies into isolated, portable "container" packages, so they run the same way in every environment.

It's used in agent security to build the [Execution Environment (Sandbox)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox). It serves a similar purpose to [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail), blocking at the hardware level any harm the code an agent runs could do to the host system.

## Doer / Executor

This is the [worker agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) that's stripped of architectural [planning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) or code-review (Reviewer) duties within the team, and just carries out the operation in the field, sticking exactly to the [plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) the [Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) agent drew up.

Its main job is to write the raw code, trigger [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) calls, and run [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) commands. It makes up the "[Maker](https://oguzerol.com/en/autonomous-ai-systems-glossary/#zapier---make)" (producer and muscle) [half](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) of the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model).

## DOM (Document Object Model)

This is the live document [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) a web browser builds when it reads an [HTML](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) page, representing every element on the page (button, text, image) as a tree structure.

A [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) works through the DOM when interacting with a page — finding buttons, filling forms, and clicking elements. It's the basic structure an agent "reads" the page through during [UI Testing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ui-testing-arayuz-testi) workflows.

## Domain

This is a website's unique address on the internet (e.g., \`example.com\`). It's routed to the server's IP address through DNS ([Domain](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dom-document-object-model) Name System).

A secure connection (HTTPS) is provided by an [SSL Certificate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ssl-sertifikasi). From an SEO standpoint, a domain's age, authority, and whether it contains a keyword can all affect rankings.

## Dynamic Context

This is a method where, instead of piling a project's massive library data into memory from end to end as [Static Context](https://oguzerol.com/en/autonomous-ai-systems-glossary/#static-context-statik-baglam), only the relevant scraps of data are pulled from [RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) based on the agent's actual need.

It works on [Lazy Loading](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lazy-loading-tembel-yukleme) logic. It doesn't waste the system's [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) capacity, and by leaving unnecessary data out, it speeds the agent up while preventing [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi) risk.

## Dynamic Tool Discovery

This is the feature where, when a new [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) gets connected to the system, the agent scans the server and reads and learns the new list of [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) and capabilities it can use on its own — without a developer having to manually introduce each tool to the agent.

Thanks to this flexibility, the agent instantly adds new [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) to its portfolio during the [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) stage, and autonomously triggers those new [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) during the [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage.

## Edges

In [State Graph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-graph-durum-grafigi) architectures like [LangGraph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#langgraph---langchain), these are the connecting paths that represent the logical transitions and data-flow direction between an agent's working stops ([Nodes](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nodes-dugumler)).

For example, the path from the "Code written" node to the "Run tests" node is an Edge. They determine the routes agents will autonomously draw based on if/else rules, and they form the system's routing mechanism.

## Embeddings

These are the numeric coordinate equivalents that machines convert the words or code blocks in a document into — arrays of numbers with hundreds of dimensions — so they can be searched "semantically" inside a [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani).

They're the basic requirement for doing [Semantic Search](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-search-anlamsal-arama). Thanks to these numbers, the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) grasps the logical closeness between pieces of text, moving [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) workflows out of keyword matching and into a semantic format.

## Environment Variables

These are key-value pairs where [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) keys, database passwords, and secret credentials are stored securely at the OS level, instead of being written directly into the source code.

They're critical for [Data Privacy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#data-privacy-veri-gizliligi) and [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim). Autonomous agents reading these variables safely during [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) connections or [Deploy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama) workflows is handled within [ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar).

## Episodic Memory

This is the experience-based [long-term memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) layer where an agent keeps, like a filmstrip, the sequence of steps ([Trajectory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#trajectory-yurutme-gecmisi)) it took to complete a complex task in the past, including the mistakes it made and how it fixed them.

Unlike [Semantic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-memory-anlamsal-bellek) data (rules and facts), Episodic Memory is entirely based on "lived experience." It keeps the agent from repeating the same mistakes on similar problems in the future.

## Evals (Evaluation / Hard Evals)

This is the process of measuring (stress-testing) the agent you've built with thousands of autonomous test scenarios — not just settling for it succeeding at a couple of simple tasks, but assessing code quality, its [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) rate, and its ability to recover from tough situations.

It's usually scored by another advanced [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) ([LLM-as-a-Judge](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-as-a-judge-hakem-llm)). In [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) architectures, an agent must pass these heavy tests before it can go to production.

## Execute

This is the final stage where the agent applies the strategy it built during the [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) stage to the physical/digital world.

In this operational step, the agent fills in the relevant [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) parameters as [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json), runs code in the [Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal), or writes a database record — carrying out an actual [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) in the outside world. Right after Execute finishes, the system moves into the [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage to evaluate the results.

## Execution Environment (Sandbox)

These are fully isolated virtual cells set up to keep the raw code and system commands autonomous agents actually run via [CLI Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-tools) from leaking into and damaging the user's host operating system.

They're provided through [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail) ([Seatbelt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sandbox-exec-seatbelt)) or [Docker](https://oguzerol.com/en/autonomous-ai-systems-glossary/#docker) containers. Backed by an [ACL (Access Control List)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi), an agent falling into an infinite loop or deleting the system is [restricted](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) at the hardware level.

## Fallback Strategy

This is when, instead of the system locking up and failing outright, the agent autonomously switches to "Plan B" (by detecting the error message) if the main [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) it's connected to crashes (an [LLM Gateway](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-gateway-model-gecidi) outage) or an external [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) stops responding.

By including [actions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) like switching to an older/cheaper backup [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) or triggering a different [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) method when an error hits, it keeps the [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) philosophy alive.

## Fast Mode (Heuristics)

This is a working mode where the agent handles routine work that doesn't need [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) — text parsing, simple translation, or [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) formatting — with much smaller, reflexive, fast [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) instead of giant ones, per the "[60/30/10 Rule](https://oguzerol.com/en/autonomous-ai-systems-glossary/#60-30-10-kurali)."

It lets the agent work off memorized rules ([Heuristics](https://oguzerol.com/en/autonomous-ai-systems-glossary/#heuristics-sezgiseller)) instead of running complex reasoning, which minimizes [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) (cost).

## Few-Shot Prompting

This is the art of giving an agent 2-3 flawless reference examples (in input-output form) of exactly what you expect, inside a [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd) file — instead of just explaining a complex task with rules alone (Zero-Shot).

Using these example patterns as a reference, the agent doesn't drift from the format you asked for. It's the most effective technique for getting [Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti) (structured [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) or custom Markdown output) and aligning the system.

## Fine-Tuning

This is the process of retraining a general-purpose [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) with data specific to one field (e.g., medical terminology, legal language, customer service) to make it a specialist in that domain.

It shouldn't be confused with [Prompt Tuning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-tuning) (prompt optimization). Fine-Tuning changes the [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) weights, while [Prompt Tuning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-tuning) only optimizes the input text. It's far more expensive in both cost and time.

## Firestore (Firebase)

This is a real-time, [serverless](https://oguzerol.com/en/autonomous-ai-systems-glossary/#serverless-sunucusuz) [NoSQL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nosql) database — commonly used in [Multi-Tenant](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-tenant-coklu-kiraci-izolasyon) or cloud-based agent projects — where agents safely sync the rules they learn autonomously, [Job Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#job-memory-gorev-hafizasi) records, and [session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) logs.

It's fast compared to local files ([JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)/TXT) and provides powerful infrastructure that lets agents instantly share [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) data across users (with isolation).

## Function Calling

This is the [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) where an [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin), instead of just turning into text for a human, reads the list of external tools ([Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri)) the system offers it, decides "I need to run this tool on the database to solve this task," and autonomously fills in the right parameters for that tool (in [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) format) and passes them to the local system.

It's the building block of [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) architectures, because it turns the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) from a passive chatbot into an active machine that can physically intervene ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) in the outside world.

## GDPR

This is the world's strictest legal framework for how personal data is collected and processed, enacted by the European Union (General Data Protection Regulation).

The mandatory Cookie Consent banner, the right to erasure, and Consent Mode compliance on websites are all requirements of this law. [KVKK](https://oguzerol.com/en/autonomous-ai-systems-glossary/#kvkk) is Turkey's equivalent regulation. Violating it leads to heavy fines.

## Gemini (Google)

This is the AI [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) family built by Google DeepMind that can process text, image, audio, and video all at once ([Multimodal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multimodal-coklu-mod)) — Gemini Pro, Ultra, Flash.

It's a direct rival to [OpenAI's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai) GPT series and [Claude](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic). Thanks to its deep integration with the Google ecosystem (Search, Ads, Analytics), it offers a strong [brain](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) alternative for [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems, especially for data-analysis and search-optimization tasks.

## GEMINI.md / AGENTS.md

This is the central rules document, sitting at a project's root directory, that holds the fundamental laws ([Workspace Rules](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workspace-rules-yerel-kurallar)) every agent working on that project must read and follow.

Where [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd) files teach agents "how to do it," these documents define "what must absolutely never be done" and the project's coding standards. Before starting a task ([Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet)), the agent loads this constitution into memory to build project-specific context.

## GEO (Generative Engine Optimization)

This is the strategy of getting your brand or content referenced by AI-powered search engines and chat assistants like [ChatGPT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chatgpt), [Gemini](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google), and Perplexity.

Unlike traditional SEO, it's not just about ranking — it focuses on the AI's "citation" criteria. [Schema Markup](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-markup-yapisal-veri), E-E-A-T authority, and clean, structured content ([Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti)) are the core [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) of this process.

## Git

This is the distributed [version](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru)-control system that timestamps every code change in a software project, making undo and parallel work possible.

The [Git Worktree](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git-worktree) feature lets autonomous agents work in isolated branches without breaking the main project. Used together with [GitHub](https://oguzerol.com/en/autonomous-ai-systems-glossary/#copilot-github-microsoft), it automates [CI/CD](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ci-cd-surekli-entegrasyon-dagitim) pipelines and [Deploy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama) workflows.

## Git Worktree

This is a [Git](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git) feature that lets you run different versions (branches) of the same project independently, in different folders, at the same time.

It's extremely valuable in autonomous systems because traditional branch-switching can confuse agents and create [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi). A [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) can safely do its work in a fully isolated worktree folder (its own [sandbox](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox)) while the main system ([Host](https://oguzerol.com/en/autonomous-ai-systems-glossary/#host-ev-sahibi---mcp)) keeps handling other operations without interruption.

## GitHub

This is a [Git](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git)-based code [hosting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hosting-barindirma) and collaboration platform. It's used worldwide as the central repository for open-source projects and enterprise software.

Autonomous agents can connect to [GitHub](https://oguzerol.com/en/autonomous-ai-systems-glossary/#copilot-github-microsoft) through an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) to autonomously run tasks like code review ([Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent)), auto-commit, opening pull requests, and triggering [CI/CD](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ci-cd-surekli-entegrasyon-dagitim).

## Global Skills

These are universal capability ([Agent Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-skills-ajan-becerileri)) packages that don't belong to just one project — they live in a central directory on the machine, like \`~/.gemini/config/[skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skills---agent-skills-beceriler)\`, and every agent on that system can use them.

Where [Local Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#local-skills-yerel-beceriler) only hold the rules of the project they live in, Global [Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skills---agent-skills-beceriler) teach an agent general-purpose capabilities (e.g., how to connect to a database, how to [deploy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama) to AWS).

## Goldfish Memory Problem

This is a [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) tendency to forget things said just a few messages ago, or critical data it found earlier.

The cause is either the [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) [limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) filling up or [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) syndrome. For agents to get past this problem, they need to connect to persistent [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) infrastructure — like [Mem0](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mem0-kalici-altyapi), [Firestore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#firestore-firebase), or local [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) files — instead of relying on momentary memory.

## Governance

These are the enterprise management policies that determine who hundreds of different autonomous agents in an organization can talk to, what data they can access, and which [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) they can't cross.

[ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) ([access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) control) and [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) (security [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar)) are the technical [implementers](https://oguzerol.com/en/autonomous-ai-systems-glossary/#doer---executor-uygulayici) of Governance. It's the most important architectural umbrella layer that prevents chaos, especially in [Multi-Tenant](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-tenant-coklu-kiraci-izolasyon) (multi-user) or financial agent systems.

## Grounding

This is when a [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), instead of relying on its own imagination ([Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon)) while answering questions or writing code, plants its feet firmly on real, [verified](https://oguzerol.com/en/autonomous-ai-systems-glossary/#verify-dogrula) data from the outside world ([RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) or [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) results).

It's the ultimate goal of the [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) cycle; it adds credibility to the system by making sure every sentence or decision the agent produces has a piece of evidence (citation) backing it up.

## Guardrails

These are the rigid **"security fence"** and safety barriers drawn around an autonomous agent.

Unlike classic [System Prompt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#system-prompt-sistem-istemi) instructions, they're not a text-based request — they include [ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) access restrictions and code-level security rules that can't be breached.

They block malicious [Prompt Injection](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-injection-zehirli-girdi) attacks and stop the agent from deleting critical data or exceeding its [Budget Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#budget-guardrails) limits.

## Hallucination

This is when an AI [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), instead of honestly saying "I don't know" about something it doesn't know, produces extremely convincing but entirely made-up and incorrect information (or non-working code).

In autonomous systems, if this fabricated code gets run directly during the [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage, it can crash the system. To prevent this:
- [Grounding](https://oguzerol.com/en/autonomous-ai-systems-glossary/#grounding-kaynaga-baglama): Tying the answer to real data
- [Curate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#curate---curation-kurasyon): Refining data beforehand
- [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture): Independent review

## Hard Evals

These are automated, mathematically provable, ruthless test batteries made up of thousands of tough scenarios — instead of measuring an agent's quality by human observation or a couple of simple examples ([Vibe Coding](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vibe-coding)).

They measure how well the system holds up under [Red Teaming](https://oguzerol.com/en/autonomous-ai-systems-glossary/#red-teaming) tests and how it escapes an infinite-loop spiral. In enterprise systems, they're the most important piece of the [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi) criteria.

## Harness

This is the abstraction (wrapper) software that wraps different brands and [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) (e.g., [OpenAI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai), [Claude](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic), [DeepSeek](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model)) and connects them to the hardware, so they can all run in a standard way inside the same [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) or structure.

It keeps the developer from having to write separate code for every [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model); by giving each [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) a standard [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling) interface, it turns it into just another gear in the larger machine.

## Heuristics

This is when agents produce reflexive, fast, cheap solutions based on predefined rules ("If x, then do y"), instead of running lengthy [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) for every decision.

It runs under [Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics). These are the shortcut, practical-intelligence steps that keep agents from burning excessive [tokens](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b) (cost) across multiple tasks.

## Hierarchical AI Agents

This is a hierarchical working [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) where agents are organized in a chain of command, like in the military.

At the top sits a [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef) who sees the big picture; below it are [Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) agents who do the planning, and at the bottom are the [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units who actually carry out the work. It's the safest [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) architecture for breaking large, complex projects into pieces and solving them.

## Host (MCP)

In [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) architecture, this is the main platform that houses the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) and the client ([Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#client-istemci---mcp)) and kicks off the process (e.g., Cursor IDE, [Claude](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic) Desktop, or [Antigravity CLI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-cli-agy)).

The Host connects to an outside [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) to offer the agent new [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) and [Resources](https://oguzerol.com/en/autonomous-ai-systems-glossary/#resources-mcp-bileseni), and it's the umbrella environment that the entire agent infrastructure runs on top of.

## Hosting

This is the server service where a website's files ([HTML](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript), [CSS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript), images) are physically stored and made available for [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) over the internet (e.g., AWS, [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) Cloud, Hetzner).

A site's speed, security, and uptime depend directly on [hosting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#host-ev-sahibi---mcp) quality. It affects [Core Web Vitals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#core-web-vitals) and [PageSpeed / Lighthouse](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pagespeed---lighthouse) scores. It's used together with a [CDN](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cdn-icerik-dagitim-agi) to optimize performance.

## Hreflang

This is the [HTML](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) tag on multilingual or multi-region websites that tells search engines "This page is for Turkish, the English version is over here."

It's critical for SEO; used incorrectly, different language versions of the same content end up competing with each other (cannibalization). It works together with the Canonical URL. Autonomous agents can scan multilingual sites for Hreflang errors and fix them.

## HITL (Human-in-the-Loop)

This is a security and oversight architecture that pauses the agent at critical, risky steps (money transfers, deleting a database, etc.) while it's running autonomously, and asks for **human approval**.

Unlike fully unsupervised [Always Proceed](https://oguzerol.com/en/autonomous-ai-systems-glossary/#always-proceed) mode, it keeps control of the system with the developer.

During its decision-making stage, the agent fires off a [Request Review](https://oguzerol.com/en/autonomous-ai-systems-glossary/#request-review) call and waits for a human to approve it.

## identity.md

This is the core character (identity) file that defines an agent's (or AI assistant's) personality, purpose, tone, and main mission.

For example, one agent might say "I'm a strict security auditor," while another says "I'm a creative marketing specialist." When the system starts up, this file is injected first into the [System Prompt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#system-prompt-sistem-istemi), shaping the agent's persona.

## Job Memory

This is the [Short-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#short-term-memory-calisma-bellegi) space where the agent keeps the state, errors hit, and stages passed for the specific task it's currently working on.

When the task finishes successfully or gets canceled, this memory is usually cleared or summarized and permanently moved into [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek). It's the answer the agent gives itself to "What was I doing?" in the moment.

## JSON Hooks

These are the agent-ecosystem equivalent of traditional webhooks. They're internal connectors that fire autonomously and kick off another process when a specific event happens in the system (e.g., a file changes, an agent hits an error).

Working integrated with [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) services, they make the system reactive and let agents communicate with each other via [A2A](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent).

## JSON Validation

This is the [validation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#verify-dogrula) layer that instantly tests whether the [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) text an agent generates during [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling) or output generation is structurally error-free (no syntax errors) and has the expected fields (e.g., "name," "age").

It's run by the [Schema Validator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-validator); if there's an error, it stops the system from breaking by autonomously sending the agent a "the [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) is broken, rewrite it" command.

## JSON-RPC 2.0

This is the stateless, quite lightweight remote procedure call protocol that forms the backbone of [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) communication.

Every message that goes back and forth between the [Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#client-istemci---mcp) and Server (requesting a [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) list, running a tool, etc.) travels entirely over this simple [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) standard, keeping the systems talking fast and stable.

## Knowledge Subagent

This is a ghost agent that reads and analyzes a project's massive codebase, documents, and past conversations in the [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) (while the user sleeps or the system is idle), converting them into [Embeddings](https://oguzerol.com/en/autonomous-ai-systems-glossary/#embeddings-yerlestirmeler) and indexing them.

Working on [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) logic, it raises the system's search quality ([Semantic Search](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-search-anlamsal-arama)) and prepares the infrastructure that lets active worker agents instantly reach the information they need.

## Knowledge Vault

This is the central storage area where agents keep the persistent knowledge they've learned ([Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek)), system logs, and user preferences in an isolated, encrypted form.

It can be kept in a cloud service like [Firestore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#firestore-firebase), or it can live entirely on your local disk (e.g., a \`.gemini\` folder). It's the physical data store that solves an agent's amnesia problem.

## Knowledge Items (KIs)

These are information packages, stored in the \`appDataDir/knowledge\` directory and containing \`metadata.json\` plus documents, that make up a project's or organization's institutional memory.

Before writing code or making a decision (during the [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) stage), an agent reads these packages to learn the project's previously made architectural decisions, known bugs (gotchas), and [SOPs](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sop-standart-opr-proseduru) (Standard Operating Procedures), joining the project from behind the [Context Wall](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-wall-baglam-duvari).

## KPI (Key Performance Indicator)

This is the numeric target metric set to measure the success of a piece of work. In marketing, values like ROAS, CTR, CPA, and Bounce Rate are all KPIs.

Autonomous agents continuously track set KPI targets during the [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage, and change strategy with [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) if a target isn't being hit. It's the basic building block of [Dashboard](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dashboard-kontrol-paneli) reports.

## KVKK

Turkey's Personal Data Protection Law. It's the legal framework Turkey put in place, for a purpose similar to [GDPR](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gdpr-kisisel-veri-koruma) (Europe), regulating how personal data is collected, processed, and stored.

The mandatory Cookie Consent banner, privacy notice, and explicit consent mechanisms on websites are requirements of this law. Autonomous agents can check KVKK compliance during site audits.

## LangGraph / LangChain

These are the most popular [orchestration](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestration-orkestrasyon) frameworks that let you build an autonomous agent's decisions and [Workflows](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar) as a node graph ([State Graph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-graph-durum-grafigi)) of logical [Nodes](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nodes-dugumler) and connections ([Edges](https://oguzerol.com/en/autonomous-ai-systems-glossary/#edges-kenarlar)), instead of plain-text code.

They keep agents from falling into infinite loops, offer rollback mechanisms, and make the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) process mathematically controllable.

## Latency

This is the wait time, in milliseconds, from when an agent ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) receives input to when it produces the first word (output) after thinking ([Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme)).

It's vital in enterprise agent architectures (e.g., voice assistants or [Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics) processes). When designing a [Compound AI System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#compound-ai-system), the right [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) is chosen by balancing speed (low latency) against intelligence (high accuracy).

## Lazy Loading

This is the strategy of dynamically pulling data from the database exactly when the agent needs it, instead of loading massive unneeded data or the [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) list into the agent's memory ([Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi)) upfront.

It prevents [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi) and multiplies performance in [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) architectures by cutting [token](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b) usage.

## Learner Agent

This is a self-training agent that analyzes the mistakes other agents in the system make (logs from the [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage) and the interventions humans make ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)), learns from them, and autonomously writes new [Local Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#local-skills-yerel-beceriler) or updates existing system prompts.

It runs the [Memory Consolidation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#memory-consolidation) process. By making the system stronger every time it makes a mistake, it makes the [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) ability permanent.

## LLM (Model / Brain)

Large Language [Model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model). This is the central brain of the agent system — the pure text understanding and generation muscle (e.g., [DeepSeek](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), GPT, [Claude](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic)).

Unlike AI assistants ([ChatGPT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chatgpt)), in autonomous systems the LLM isn't a standalone product — it's just one of hundreds of components inside a [Compound AI System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#compound-ai-system). The agent uses the LLM brain to do [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) and trigger [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri).

## LLM Gateway

This is the proxy/server layer an agent passes through — instead of connecting directly to the AI [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) behind it ([OpenAI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai), [Anthropic](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic), etc.) — that manages [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) keys, [Rate Limiting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rate-limiting-hiz-kisitlamasi), and load balancing (e.g., LiteLLM).

It's the center of the [Fallback Strategy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fallback-strategy-yedek-plan) implementation. When a [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) crashes or stops responding, the Gateway instantly shifts traffic to another [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), letting the agent keep working uninterrupted instead of falling into an infinite-loop spiral.

## LLM-as-a-Judge

This is another independent, usually smarter (e.g., top-tier reasoning) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) used to score and audit the output an agent produces during [Evals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#evals-evaluation---hard-evals) (stress tests) or quality-control stages, without needing human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) intervention.

It serves as the "Checker" in the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model). It runs the code an agent writes and scores the output out of 10, maximizing automation speed.

## Local Skills

These are expertise (know-how) packages that only live inside the folder of the project they belong to (e.g., \`.agents/[skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skills---agent-skills-beceriler)\`), version-controlled with [Git Worktree](https://oguzerol.com/en/autonomous-ai-systems-glossary/#git-worktree), holding architectural rules and [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) usage patterns specific to that project.

Unlike [Global Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#global-skills-kuresel-beceriler) (system-wide skills), Local [Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skills---agent-skills-beceriler) are isolated. They give an agent targeted, permanent [Procedural Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#procedural-memory-prosedurel-hafiza), like "How do you write [React](https://oguzerol.com/en/autonomous-ai-systems-glossary/#react-reasoning--acting) components on this specific project?"

## Long-Term Memory

This is the **"permanent memory store"** that carries an agent's experience from past sessions, user preferences, and institutional knowledge forward into the future.

It gets past the limitation of [Short-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#short-term-memory-calisma-bellegi), which gets wiped at the end of a session.

It stores data in a [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani) and feeds [Semantic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-memory-anlamsal-bellek) and [Procedural Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#procedural-memory-prosedurel-hafiza) structures, letting the agent learn over time ([Learner Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#learner-agent-ogrenen-ajan)).

## Loop Engineering

This is the discipline of designing unbreakable [workflows](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar) where agents trigger and correct themselves inside an [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) until they reach the goal, instead of working off one-off commands (classic [Prompt Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-engineering)).

The goal is to build solid [Fallback Strategy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fallback-strategy-yedek-plan) mechanisms to keep agents from crashing, and to set firm [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi) stop conditions that keep them from falling into an infinite-loop spiral.

## Lost in the Middle

This is the ailment where an agent ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) — once thousands of lines of code or text have piled up inside a massive-capacity [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) — remembers the very start and very end of the document but overlooks and skips critical [sections](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) sitting right in the middle.

To prevent this memory poisoning ([Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi)), instead of dumping everything into the system directly, it should be given [curated](https://oguzerol.com/en/autonomous-ai-systems-glossary/#curate---curation-kurasyon) summaries, and data should be broken up with [Semantic Chunking](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-chunking).

## Maker-Checker Architecture

This is the autonomous, asynchronous control loop between the [Implementer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementer-agent-uygulayici) that actually produces the work ([Maker](https://oguzerol.com/en/autonomous-ai-systems-glossary/#zapier---make)) and the [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) that ruthlessly audits the resulting output (Checker).

Since having a human review every piece of code ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) would create a bottleneck, this architecture lets code and text get produced at high quality and error-free through cross [Red Teaming](https://oguzerol.com/en/autonomous-ai-systems-glossary/#red-teaming) tests, without needing human intervention.

## MCP (Model Context Protocol)

This is the universal **"plug-and-play USB cable"** standard that lets AI models talk to databases, file systems, and external services.

Unlike traditional [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) or custom integration solutions, it removes the need to code a separate [API Connector](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api-connector) for every platform.

It manages the capabilities offered to the AI as [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-mcp-bileseni), data as [Resources](https://oguzerol.com/en/autonomous-ai-systems-glossary/#resources-mcp-bileseni), and command templates as [Prompts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompts-mcp-bileseni) — 3 core [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) components.

## MCP Client

In [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) architecture, this is the bridge interface that runs inside the [Host](https://oguzerol.com/en/autonomous-ai-systems-glossary/#host-ev-sahibi---mcp) application (e.g., Cursor or [Antigravity IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravity-ide)) and kicks off communication between the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) and the external server.

It receives the [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling) requests the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) makes in natural language or simple [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json), converts them into standard [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)-RPC packets, and forwards them to the [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu), letting the agent exchange data with the outside world.

## MCP Server

This is the service (server) layer that opens up an external database, a [GitHub](https://oguzerol.com/en/autonomous-ai-systems-glossary/#copilot-github-microsoft) repository, or a local folder to the outside world by translating it into a standard format the agent ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) can understand.

It offers the agent 3 core components: [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) ([actions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)), [Resources](https://oguzerol.com/en/autonomous-ai-systems-glossary/#resources-mcp-bileseni) (readable data), and [Prompts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompts-mcp-bileseni) (templates). The agent connects to this server through an [MCP Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-client-mcp-istemcisi) and autonomously manages external systems within its permissions ([ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi)).

## Mem0

This is an external [long-term memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) database solution built specifically to solve agents' amnesia problem ([Goldfish Memory Problem](https://oguzerol.com/en/autonomous-ai-systems-glossary/#goldfish-memory-problem)).

Even after [sessions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) ([Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek)) close, it stores the important past conversations an agent had with a user, their preferences, and [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek) logs. It gives the agent a consistent [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) experience, letting it remember the user in every new [session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum).

## Memory Consolidation

This is the process of an agent analyzing and summarizing the complex error logs, conversations, and code attempts that built up in [Job Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#job-memory-gorev-hafizasi) throughout the day, and turning them into clean, permanent institutional knowledge ([Knowledge Items](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis)) at the end of the day (or when the task finishes).

It prevents [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi) and ensures the agent only carries distilled experience into [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek).

## Meta Tags

These are the tags in an [HTML](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) page's \`\` section that tell search engines and social media platforms about the page.
- Title Tag: The page title (the blue link in [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) results)
- Meta Description: The page description (the gray text in [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) results)
- Canonical: The page's true address

They directly affect SEO rankings and CTR. Autonomous agents can spot and fix missing or poorly written meta tags.

## Metadata Tagging

This is the process of adding invisible identity ([JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)) tags — like "date, author, topic, expiry" — to files agents create, especially [Knowledge Items](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) packages.

During [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) and [Semantic Search](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-search-anlamsal-arama), it lets the agent filter down to just the data with relevant tags instead of getting lost among billions of records. It's the most fundamental [indexing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#indexing-indeksleme) rule that keeps large [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) projects organized.

## Multi-Agent System (MAS)

This is the **"team and swarm architecture"** where multiple specialist agents work together to accomplish a complex task.

Instead of loading every task onto one giant agent, it divides labor among specialist [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units (coder, tester, researcher) under the leadership of an [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan).

Agents communicate with each other via the [A2A (Agent-to-Agent)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent) protocol.

## Multi-Tenant (Isolation)

This is the architecture where the data of hundreds of different users or companies using the same agentic infrastructure (e.g., a [Telegram bot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#telegram-bot) or a SaaS panel) is kept in fully isolated cells in databases like [Firestore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#firestore-firebase), never mixing together.

It's critical for [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim); thanks to [Identity Propagation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#identity-propagation-kimlik-akt), an agent reading Customer A's file can never accidentally reach Customer B's data (enforced by [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar)).

## Multimodal

This is an AI [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) ability to understand and process not just text, but different data types at once — image ([Vision](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme)), audio, and video.

Current [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) like GPT and [Gemini](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) are multimodal. Multimodal ability is used when a [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) analyzes screenshots, spots visual bugs during [UI Testing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ui-testing-arayuz-testi), or evaluates ad creative.

## n8n

This is an open-source workflow automation platform you can host on your own server (self-[hosted](https://oguzerol.com/en/autonomous-ai-systems-glossary/#host-ev-sahibi---mcp)). It's an alternative to [Zapier / Make](https://oguzerol.com/en/autonomous-ai-systems-glossary/#zapier---make), but offers advantages in [data privacy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#data-privacy-veri-gizliligi) and cost.

Autonomous agents can trigger n8n [workflows](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar) through an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) or Webhook to run multi-step automations like CRM updates, email automation, or report generation.

## Narrow AI

This is the concept describing today's general state of AI — able to perform only a specific task (e.g., playing chess, writing code, generating images) at or above human level, but unable to understand anything outside that.

It's the exact opposite of the hypothetical do-everything level of [AGI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agi-genel-yapay-zeka). Agentic systems aim to get past these narrow [limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) by combining different Narrow AI specialists under a single [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan).

## Nodes

In architectures like [LangGraph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#langgraph---langchain) or a [State Graph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-graph-durum-grafigi), these are the independent working stops an agent passes through, each carrying out a function (e.g., "search the web," "test the code," "produce the answer").

Nodes are connected to each other by [Edges](https://oguzerol.com/en/autonomous-ai-systems-glossary/#edges-kenarlar). When the agent finishes its work at one node, it moves the data it produced (State) to the next node, advancing the autonomous [workflow](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar).

## NoSQL

Unlike traditional table-based (relational) [SQL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sql) databases, this is a database family that stores data as flexible-schema ([JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)-like) documents (e.g., MongoDB, [Firestore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#firestore-firebase), Redis).

It's well suited for autonomous agents to quickly read and write [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) and [Job Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#job-memory-gorev-hafizasi) data. In [Multi-Tenant](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-tenant-coklu-kiraci-izolasyon) architectures, user data is kept isolated in NoSQL collections.

## Notification System

This is the mechanism autonomous agents use to send instant notifications to a human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) when a critical event happens (an error, a task finishing, a budget overrun).

It works through channels like a [Telegram Bot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#telegram-bot), [Slack Integration](https://oguzerol.com/en/autonomous-ai-systems-glossary/#slack-integration), email, or push notifications. Integrated with [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) services and [JSON Hooks](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-hooks-json-kancalari) triggers, it enables autonomous monitoring.

## nsjail

This is a Linux-based, extremely strict hardware-level [Execution Environment](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox) (Sandbox / isolated cell) tool used to keep the raw code autonomous agents actually run through [CLI Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-tools) ([Terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) commands) from leaking into and damaging the user's host operating system.

Working on [ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) ([access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) control) logic, it locks down the agent's processor, memory, and network [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim). It's the ultimate fuse for system security.

## Observe

This is the **"sense and perceive"** phase where the agent reads the system's new state, error output, and environment responses after carrying out an [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)).

It's the control mechanism of the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) cycle; it feeds the incoming output into the [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) stage, letting the agent update its next [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) step.

For example, when a command that's run throws an error, the agent observes that state and kicks off the [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) process.

## Open Source vs Closed Source

This is whether an AI [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) or piece of software's source code is open to everyone (Open Source) or kept private by the company (Closed Source).
- Open Source: [DeepSeek](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), LLaMA, Mistral — can be run on your own server, giving you a [data privacy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#data-privacy-veri-gizliligi) advantage.
- Closed Source: [OpenAI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai) GPT, [Claude](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic), [Gemini](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) — used through an [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api), data goes to the company's servers.

It's a critical choice in [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim) and [Data Privacy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#data-privacy-veri-gizliligi) decisions.

## OpenAI

This is the pioneering company in autonomous AI that built [ChatGPT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chatgpt) and the GPT family of [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model).
- Standard [Models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) (GPT series): fast, low-cost [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), preferred for both [Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics) and general tasks.
- [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) [Models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model): deep-[reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) that solve complex problems by running [Chain of Thought](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chain-of-thought-cot) for minutes at a time.

Rivals: [Claude (Anthropic)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic), [Gemini (Google)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google), [DeepSeek](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) (open source). Because the top-tier [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) in particular carry high cost and processing time, under the [60/30/10 Rule](https://oguzerol.com/en/autonomous-ai-systems-glossary/#60-30-10-kurali) they're usually reserved for only the most critical [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) tasks (architectural decisions).

## Orchestration

This is the process of managing dozens of otherwise disorganized, independently running agents, [APIs](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api), and tools ([Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri)) in harmony — like a symphony orchestra — around a defined [workflow](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar) and [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim) rule, without creating chaos.

It's the main feature that sets [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems apart from simple chatbots. It manages errors ([Fallback Strategy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fallback-strategy-yedek-plan)), distributes tasks, and keeps the system whole.

## Orchestrator

In multi-agent systems ([Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru)), this is the **"main manager"** agent that runs the orchestra and breaks complex tasks into sub-parts.

It distributes work to [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units, routes processes with [Agent Routing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-routing), and synthesizes the results that come back.

It's the heart of the [Orchestrator-Worker Pattern](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-worker-pattern) architecture; when needed, it gets [HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop) approval from a human to oversee the process moving forward.

## Orchestrator-Worker Pattern

This is the classic, hierarchical, centralized agent architecture template, with a Conductor ([Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan)) who sees the big picture at the top, and Workers ([Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan)) below it who actually carry out specific jobs (development, design, testing).

It's the most commonly chosen structure for avoiding chaos in large projects. Even though workers communicate with each other via [A2A](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent), the final decision and approval always comes from the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) (or a human, via [HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)).

## Output Format

This is when the response an agent ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) produces comes in a structured data format (e.g., [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json), XML, Markdown) that external systems or other agents can read and run directly, instead of random plain text.

It's the foundation of [Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti) requirements. Using validation mechanisms like [Few-Shot Prompting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#few-shot-prompting) and [Pydantic](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pydantic) ([JSON Validation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-validation-dogrulama)), it keeps the agent from drifting out of format ([Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon)) and ensures [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling) runs error-free.

## PageSpeed / Lighthouse

These are [Google's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) free performance testing [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) that measure a web page's load speed, accessibility, and SEO compliance.

They report [Core Web Vitals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#core-web-vitals) (LCP, INP, CLS) metrics. Autonomous agents can autonomously run Lighthouse tests through a [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) and suggest [CDN](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cdn-icerik-dagitim-agi), image compression, or code optimizations for low-scoring areas.

## Plan

This is the second of the 4 core stages inside the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) ([Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet), Plan, Reason, [Act](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)). It's the stage where, after [observing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) the environment, the agent breaks a large, complex task into small, manageable sub-tasks (steps).

The agent doesn't move straight to [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) at this stage. [Deliberative](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deliberative---planning-agents) Agents ([Planner Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#planner-agent-planlamaci-ajan)) in particular map out the system's route at this stage by preparing a detailed blueprint (architecture) or [Task Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#task-plan-taskmd) (task list).

## Planner Agent

In hierarchical ([Hierarchical AI Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hierarchical-ai-agents)) systems, this is the analytical specialist agent focused purely on building strategy, instead of writing code directly ([Doer](https://oguzerol.com/en/autonomous-ai-systems-glossary/#doer---executor-uygulayici)/[Executor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#doer---executor-uygulayici)) or debugging.

Its job is to break the main request it gets from the [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef) into pieces, decide which [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) to use, and produce an [Implementation Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementation-plan). It submits the [plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) it produces for approval ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) and, once approved, hands it off to [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units to carry out.

## Pragmatist Agent

This is a results-focused agent that clashes with the [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) hunting for perfection or overly academic code standards inside [Debate Loops](https://oguzerol.com/en/autonomous-ai-systems-glossary/#debate-loops-tartisma-odalari), keeping the system focused on the "fastest, working" solution instead.

Especially when [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi) criteria are flexible, it stops the system from falling into an infinite-loop spiral and burning too much [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi); it pushes the good-enough solution through.

## Presenter Agent

Unlike lower-level agents that work in the [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) and produce complex database logs and [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) formats, this is the interface agent that turns raw results into a visual, interactive [Artifact](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar) or [Walkthrough](https://oguzerol.com/en/autonomous-ai-systems-glossary/#walkthrough-yurutme-raporu) (report) the end user (a human) can understand.

Using the [Progressive Disclosure](https://oguzerol.com/en/autonomous-ai-systems-glossary/#progressive-disclosure) principle, it doesn't drown the user in information — it just presents the clean summary needed for approval ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)).

## Principle of Least Privilege

This is the rule that an agent or a [Sidecar](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sidecars-yan-surecler) microservice should only ever be given the "minimum level" of permission it needs to do its job.

It's the backbone of [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim) and system security. If an agent only needs to read a folder, it's never given write or delete permission for it through the [ACL (Access Control List)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi). This way, if a [Prompt Injection](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-injection-zehirli-girdi) attack succeeds, the system as a whole isn't [harmed](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme).

## Privilege Boundaries

These are the invisible walls that separate the permission and [access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) limits of different agents in a system (e.g., a Customer Service Agent vs. a Database Admin Agent) or different projects ([Multi-Tenant](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-tenant-coklu-kiraci-izolasyon)), both at the hardware and software level.

They're strictly managed by the [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist). An agent that tries to violate these boundaries is instantly blocked by [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) or [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail) ([Sandbox](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox)), stopped with a [Segmentation Fault](https://oguzerol.com/en/autonomous-ai-systems-glossary/#segmentation-fault-bellek-tasmasi)-style emergency hard stop.

## Procedural Memory

This is the persistent expertise memory where an agent keeps the steps (the way of doing things) for specific work, like "How do you create a [React](https://oguzerol.com/en/autonomous-ai-systems-glossary/#react-reasoning--acting) component?" or "How do you [deploy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deploy-yayinlama) to AWS?"

It's loaded into the agent through [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd) files or [Local Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#local-skills-yerel-beceriler) packages. Unlike [Semantic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-memory-anlamsal-bellek) (facts) or [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek) (past experiences), it holds task standards that are entirely [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)-based and reflexive, like muscle memory.

## Progressive Disclosure

Instead of dumping a massive analysis result on a human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) all at once and breaking the [Context Wall](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-wall-baglam-duvari) illusion (drowning the user in information), this is when agents present information step by step, through an interface that only expands as needed.

It's usually done with [Artifacts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar) or [Walkthrough](https://oguzerol.com/en/autonomous-ai-systems-glossary/#walkthrough-yurutme-raporu) documents. It's a critical UX principle that reduces the cognitive load on a human auditing ([Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle)) the system.

## project-context.md

This is the static orientation file at a project's root directory that summarizes the project's overall state, architecture, directory structure, and what the folders are for, for the agent ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) on first entry.

Unlike [GEMINI.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#geminimd---agentsmd) (rules) or [identity.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#identitymd) (personality) files, it exists entirely to answer "What is this project, what's where?" and keeps the agent from flying blind during the [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) stage.

## Prompt Caching

This is a technology where, instead of a [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) re-reading (processing) fixed input that repeats on every request — the giant [System Prompt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#system-prompt-sistem-istemi), project-specific rules ([AGENTS.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#geminimd---agentsmd)), old logs — it keeps this text in memory (Cache) and only processes the newly added [parts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis).

It's a revolutionary optimization method in agentic systems that cuts [Latency](https://oguzerol.com/en/autonomous-ai-systems-glossary/#latency-gecikme-suresi) and [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) (costs) by up to 90%.

## Prompt Contract

This is the unwritten agreement between an agent and a developer. It's where the developer rigidly defines, in the main prompt, the specific [Output Format](https://oguzerol.com/en/autonomous-ai-systems-glossary/#output-format-cikti-formati) they expect from the agent, the [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) rules it must follow, and the [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) permissions it will use.

It's backed technically by a [Schema Validator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-validator) and [Pydantic](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pydantic). The moment the agent steps outside the contract, an error is thrown ([Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle)) and the loop is stopped.

## Prompt Engineering

Instead of the art of getting the best answer out of a [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) with magic words, this is the mathematical and structural engineering of designing commands (prompts) — in agentic ([Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai)) systems — that maximize the [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) capacity and keep it faithful to a format using [Few-Shot Prompting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#few-shot-prompting).

In modern systems, it's gradually giving way to autonomous [Loop Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#loop-engineering-dongu-muh) and dynamic [RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) ([Context Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-engineering)) designs.

## Prompt Tuning

Instead of a developer writing long instructions ([Prompt Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-engineering)) every single time, this is optimizing the base [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) [model's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) native language and reflexes for that specific job (e.g., debugging) with low-cost, permanent weight updates (soft tuning or LoRA).

It saves the agent from heavy [token](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b) costs, letting it operate at enterprise standard even with much smaller prompts inside [Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics).

## Prompt Injection

This is a cyberattack method where malicious users or manipulative data sources try to trick an agent into breaking its original [System Prompt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#system-prompt-sistem-istemi) rules.

It's similar to SQL Injection on classic websites; it tries to leak hidden instructions into the agent or get it to run unauthorized commands.

Strict [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) and system-level [ACL (Access Control List)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) rules are applied to protect the system against these kinds of attacks.

## Prompts (MCP Component)

This is one of the 3 core components ([Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri), [Resources](https://oguzerol.com/en/autonomous-ai-systems-glossary/#resources-mcp-bileseni), Prompts) an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) offers to an agent ([Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#client-istemci---mcp)).

These are pre-built [templates](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar) that explain how the agent should do a task on that server. The developer doesn't need to write out instructions like "how to search for a customer in Salesforce" for the agent; the [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) autonomously hands this standard Prompt (template) to the agent directly during [Dynamic Tool Discovery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-tool-discovery).

## Protobuf (.pb)

Instead of sub-components in autonomous agent systems (e.g., [Sidecars](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sidecars-yan-surecler) or [Worker Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan)) sending massive [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) files back and forth and slowing the system down, this is a messaging standard (gRPC infrastructure) that converts data into a compressed binary format a computer can read much faster.

By solving [Latency](https://oguzerol.com/en/autonomous-ai-systems-glossary/#latency-gecikme-suresi) problems, it becomes the core data carrier for [A2A](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent) communication in high-performance [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) architectures.

## Pydantic

This is a validation library that automatically checks the free-form [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) output an agent ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) produces against predefined, strict Python classes (data schemas) — and warns the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), forcing a fix, if the wrong type of data comes through (e.g., writing "thirty" instead of an age number).

It takes on the [Schema Validator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-validator) role to guarantee [Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti) and keep the system from breaking during [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling).

## RAG (Retrieval-Augmented Generation)

This is an **"open-book exam"** technique that steers a model toward the right answer by pulling relevant information from external data sources (a [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani)), instead of training the AI model from scratch.

It stops the model from [Hallucinating](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) and verifies information through [Grounding](https://oguzerol.com/en/autonomous-ai-systems-glossary/#grounding-kaynaga-baglama).

Combined with autonomous loops, it turns into the smarter [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) structure.

## Rate Limiting

This is the shield that stops autonomous agents, when they fall into an uncontrolled infinite loop, from hammering external systems ([APIs](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api), databases) or the main [LLM Gateway](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-gateway-model-gecidi) connection thousands of times in seconds — locking it up (DDoS) or burning through the budget.

It works together with [Budget Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#budget-guardrails), keeping the system stable by imposing hardware-level limits on the agent, like "you can make a maximum of 10 requests per minute."

## Reach

This is the number of unique people an ad or piece of content reaches. It shouldn't be confused with Impressions; if one person sees an ad 3 times, Reach is 1 and Impressions is 3.

It's the most important [KPI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#kpi-anahtar-performans-gostergesi) metric in brand-awareness campaigns. Detailed Reach reports are provided on the Meta Ads platform.

## ReAct (Reasoning + Acting)

This is the core paradigm that separates agents from traditional [LLMs](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) that "just think and produce code." It's the flawless, repeating cycle between the [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) stage, where the [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) internally plans and reasons, and the [Acting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage, where it directly intervenes in the outside world with [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri).

It's the ancestor of modern [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) designs, giving an agent the ability to both think like a philosopher and act like a worker.

## Reactive Agents

This is a reflexive agent profile that, when a problem comes up, jumps into action instantly, fixes and retries as errors come in, and works on fast reflexes — instead of pulling back to write long [Implementation Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementation-plan) documents like [Deliberative](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deliberative---planning-agents) ([Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)) Agents.

While the risk of mistakes is higher on big architectural designs, they provide unmatched speed for simple code fixes, formatting, and quick lookups ([Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics)).

## Reasoning

This is the **"thinking phase"** where an AI model builds logical steps in the background and reasons through them before producing an answer.

Instead of jumping straight to an answer and [Hallucinating](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon), it breaks the problem into pieces and analyzes it using [Chain of Thought (CoT)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chain-of-thought-cot) or [Tree of Thoughts (ToT)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tree-of-thoughts-tot) methods.

It's the bridge between [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) and [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) inside the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu).

## Red Teaming

This is the process of ruthlessly stress-testing your own supposedly flawless agent system from the inside — by having other agents loaded with [Contrarian Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#contrarian-agent-aykiri-ses) or [Adversary Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#adversary-agent-dusman) roles attack it — to see whether [Prompt Injection](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-injection-zehirli-girdi) holes exist and whether [Privilege Boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar) can be broken.

It's the most critical [Hard Evals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#evals-evaluation---hard-evals) security stage required before an [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) architecture can go safely into production.

## Redirect

This is when a URL automatically forwards to another URL.
- 301 (Permanent): The page has moved, and SEO strength carries over to the new address.
- 302 (Temporary): The page is temporarily somewhere else, and SEO strength doesn't transfer.

Incorrect redirects cause [Crawling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#crawling-tarama) errors and lost SEO rankings. Autonomous agents can scan for broken links and bad redirect chains and fix them.

## Reducers

In [LangGraph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#langgraph---langchain) or [State Graph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-graph-durum-grafigi) structures, these are functions that merge multiple pieces of data coming from different [Nodes](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nodes-dugumler) or parallel-running agents (e.g., different design ideas or data batches) into a single, coherent, clean structure (State).

Supporting [Memory Consolidation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#memory-consolidation) and [Curate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#curate---curation-kurasyon) processes, they prevent wasted agent memory and keep the system from falling into [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) syndrome.

## Reflective Memory

This is a deep-learning memory where the agent stores the lessons it produces internally ([Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme)) about why it made a past mistake — working like "I tried this code and it crashed because library X was missing, next time install X first."

Where [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek) only keeps "what happened," Reflective Memory keeps "what lesson was learned from it." It's the most valuable type of [Knowledge Item](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) (KI) for letting an agent autonomously master its craft (Self-Improvement).

## Request Review

In [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems, this is the [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) where an agent pauses the process and requests approval from a human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) or a review from a senior conductor ([Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan)), either at the end of a task ([Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi)) or before triggering a dangerous [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api).

It's the system's safety brake when [Always Proceed](https://oguzerol.com/en/autonomous-ai-systems-glossary/#always-proceed) mode isn't active. A human reviews the code, tells the autonomous agent "fix this part," and sends the agent back to the [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage to keep it from making a mistake.

## Resolver Agent

Inside a [Multi-Agent System (MAS)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru), this is a specialist (worker) agent with strong problem-solving skills, focused purely on fixing a bug or a broken build, instead of writing new code.

For example, if the Implementer agent's code doesn't compile, the [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef) doesn't step in — instead, the Resolver agent is called directly through [A2A (Agent-to-Agent)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent) communication. The Resolver analyzes the error, [applies](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) the patch, and gets the system back on its feet.

## Resources (MCP Component)

This is one of the 3 core components an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) offers to an agent ([Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#client-istemci---mcp)) — the others being [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) and [Prompts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompts-mcp-bileseni).

It represents external data sources the agent can only read (read-only), never modify. For example, a database schema, a company's [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) documentation, or log files are all offered to the agent as a "Resource." By pulling these resources into its [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi), the agent writes more consistent, [grounded](https://oguzerol.com/en/autonomous-ai-systems-glossary/#grounding-kaynaga-baglama) code.

## Responsive Design

This is a design approach that lets a website automatically adapt and look right across different screen sizes — [desktop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru), tablet, and mobile phone.

It directly affects [Core Web Vitals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#core-web-vitals) and SEO rankings. Autonomous agents produce responsive code using [CSS / HTML / JavaScript](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) and run [UI Testing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ui-testing-arayuz-testi) across different screen sizes with a [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser).

## Reverse Prompting

This is a technique where, instead of a human telling the agent what to do, the agent [steers](https://oguzerol.com/en/autonomous-ai-systems-glossary/#redirect-yonlendirme) or interviews the developer by asking, "What's the best way for me to do this — give me info X and Y."

It's used when requirements are unclear (during the [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) stage). The agent takes the initiative autonomously, filling in the missing pieces of the [Prompt Contract](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-contract-ajan-sozlesmesi) with human approval ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)).

## Reviewer Agent

This is an independent quality-control agent built purely for auditing and review, taking on the "Checker" role in the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model).

When the [Implementer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementer-agent-uygulayici) finishes its work, it looks at this output with an unbiased eye, analyzing code standards, performance, and security vulnerabilities. If it finds an error, it sends autonomous feedback for a fix; if it approves, the code gets merged.

## Robots.txt

This is the directive file, sitting in a website's root directory, that tells search engine bots ([Googlebot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google)) which pages to crawl ([Crawling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#crawling-tarama)) and which to skip.

Misconfiguring it can leave important pages out of [Indexing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#indexing-indeksleme). It works together with the [Sitemap](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sitemap-site-haritasi). Autonomous agents can analyze robots.txt errors and suggest fixes.

## Runtime

Unlike the static moments when code or an agent sits frozen and waiting, this is the dynamic, live execution period when the system is actively processing data, making decisions, and running [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri).

The agent's memory ([Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek)) is active for this entire period. Unexpected errors or [Latency](https://oguzerol.com/en/autonomous-ai-systems-glossary/#latency-gecikme-suresi) issues that come up during Runtime can only be overcome with a strong [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) and a flexible [Fallback Strategy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fallback-strategy-yedek-plan).

## Sandbox

These are fully isolated virtual working cells set up to keep the raw code autonomous agents run through [CLI Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-tools) from leaking into and damaging the user's host operating system.

They're provided through [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail) (Linux), [sandbox-exec](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sandbox-exec-seatbelt) (macOS), or [Docker](https://oguzerol.com/en/autonomous-ai-systems-glossary/#docker) containers. Backed by an [ACL (Access Control List)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi), an agent falling into an infinite loop or deleting the system is [restricted](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) at the hardware level. It's the common name for the [Execution Environment](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox) concept.

## sandbox-exec (Seatbelt)

This is an isolation ([Sandbox](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox)) mechanism specific to Apple (macOS) systems that [restricts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis), at the kernel level, the [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) commands an agent runs from [reaching](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) OS folders (e.g., Documents, Desktop) or network ports.

It's the Mac equivalent of the [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail) tool on Linux. In [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) systems, it's an [ACL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#acl-erisim-kontrol-listesi) barrier that uses hardware-level force to stop an agent from accidentally wiping your local machine.

## Scheduled Tasks

This is scheduling infrastructure that lets agents wake up at set intervals (e.g., every hour, every night) and carry out autonomous tasks independently of a human. Also known as [Cron](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cron-zamanlanmis-gorevler).

It's managed at the [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) level. Working integrated with [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) (background) services, it enables jobs like campaign updates, backups, or system maintenance to run autonomously, 24/7.

## Schema Markup

This is code, in machine-readable [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)-LD format, that tells [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) what your page's content actually is (a product, a recipe, an FAQ, an event, a business, etc.).

It's what makes "rich results" (Rich Snippets) — star ratings, pricing info, FAQs — show up in search results (SERP). It's a critical component of SEO and [GEO](https://oguzerol.com/en/autonomous-ai-systems-glossary/#geo-generative-engine-optimization) (AI search) strategies.

## Schema Validator

This is technical shield software (e.g., [Pydantic](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pydantic)) that autonomously checks, on every operation, whether the [Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti) (structured [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)) an agent produces matches the rule set (schema) the system expects.

If the agent returns text instead of a number for the "Age" field ([Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon)), the Validator instantly throws an error, warns the agent, and gets it to fix its output ([Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma)) before anything breaks.

## SDLC (Software Development Life Cycle)

This covers the entire process from a piece of software's birth ([planning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)) through coding, testing, and going live (Production).

In [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) architectures, the goal isn't just to use an agent as a coder — it's to run the entire SDLC cycle end to end autonomously, without human hands, using [Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla), Implementer, and Reviewer agents in a [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru).

## Segmentation Fault

This is when an agent, trying to read an unbounded folder (e.g., \`node_modules\`), drowns under massive amounts of data, exceeds its [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) capacity, and the system suddenly crashes (a hard crash).

This hardware-level crash may not even be recoverable at the [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) level. To prevent it, blacklist filters like \`[.antigravityignore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#antigravityignore)\` or hardware-level [Cost Control Limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cost-control-limits) must be [applied](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula).

## Self-Correction

This is an agent's ability to [observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) an error it hits after an [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) and fix its own code or plan without human intervention.

Unlike traditional software, which crashes the moment it hits an error, this is the greatest strength of autonomous systems.

It works on the Evaluator-Optimizer Pattern and [Loop Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#loop-engineering-dongu-muh) principles.

## Self-Describing API

This is a smart structure where an external [application's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) ([API's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api)) own info — "What am I for, and what parameters do you need to write into [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) to pull data from me" — is built right into it as a schema.

Tools offered through an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) are usually Self-Describing. This lets the agent autonomously learn the [API's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) usage manual in seconds during [Dynamic Tool Discovery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-tool-discovery).

## Semantic Chunking

This is the process of splitting a massive PDF document or code file into smart pieces while preserving paragraphs' "unity of meaning" (headings, code blocks), instead of splitting by a random word count.

It determines the quality of the data going into a [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani). Badly chunked data causes an agent's [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) searches to turn up irrelevant results (noise) and leads to [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) syndrome.

## Semantic Memory

This is the library-style memory where an agent stores general facts, definitions, and universal programming standards — things like "the capital of Turkey is Ankara" or "[React](https://oguzerol.com/en/autonomous-ai-systems-glossary/#react-reasoning--acting) uses functional components instead of class components."

Unlike experience-based [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek), it holds objective facts. [Global Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#global-skills-kuresel-beceriler) and a base [LLM's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) pre-trained knowledge both fall into this memory type.

## Semantic Routing

This is the technology that [routes](https://oguzerol.com/en/autonomous-ai-systems-glossary/#redirect-yonlendirme) an incoming user request to the most accurate specialist agent by grasping the intent and meaning ([Embeddings](https://oguzerol.com/en/autonomous-ai-systems-glossary/#embeddings-yerlestirmeler)) behind the text, instead of simple if/else keyword matching.

For example, even if a request like "My site is loading really slowly" doesn't contain the word "speed," the system, via the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan), autonomously understands this problem belongs to the "Performance and Database Optimization" ([Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan)) team.

## Semantic Search

This is a technique for searching a database (e.g., a [Vector Database](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vector-database-vektor-veritabani)) based on the mathematical closeness of concepts ([Embeddings](https://oguzerol.com/en/autonomous-ai-systems-glossary/#embeddings-yerlestirmeler)), rather than exact letter-for-letter word matches.

Search for "dog" and it can find a document about a "Golden Retriever" even if the word "dog" never appears in it. It's the core engine that raises the retrieval quality and [Grounding](https://oguzerol.com/en/autonomous-ai-systems-glossary/#grounding-kaynaga-baglama) accuracy of [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) systems to the next level.

## Sequential Agents

Inside a [Multi-Agent System (MAS)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru), this is a hierarchical [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) where agents work in sequence, like an assembly line, instead of working fully independently of one another.

For example, Agent A does the research; once it's done, Agent B codes the findings; once that's done, Agent C tests it. Compared to [Debate Loops](https://oguzerol.com/en/autonomous-ai-systems-glossary/#debate-loops-tartisma-odalari) (simultaneous discussion), this gives a more deterministic [workflow](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar) that's easier to manage (and trace).

## Serverless

This is a [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) where the developer just writes the code and hands it off to a cloud provider (AWS Lambda, [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) [Cloud Functions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cloud-functions-sunucusuz-fonksiyon)), without managing a physical server. The code only runs when triggered, and you're billed only for what you use.

It's ideal for lightweight tasks autonomous agents kick off via Webhook triggers. It's the cloud-based alternative to [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) processes.

## Session

In Google Analytics (GA4), this is the core measurement unit covering the time between a user's arrival on and departure from your site.

Multiple Events (page views, clicks, scrolls) happen within a single Session. A session automatically closes after 30 minutes of inactivity. It's the foundation of Bounce Rate and Conversion Tracking calculations.

## Short-Term Memory

This is the **"momentary working memory"** an agent uses during its currently active task, holding the conversation history and temporary steps.

Unlike [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek), it's not permanent — it gets wiped when the [Session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) ends, and its capacity is limited by the [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi).

Together with [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek) and [Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek) components, it lets the agent do moment-to-moment [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme).

## Sidecars

These are small microservices that run outside the core [Agent OS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agent-os-ajan-isletim-sist) or [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan), attached to it (like a sidecar motorcycle) but running as their own separate [Daemon](https://oguzerol.com/en/autonomous-ai-systems-glossary/#daemon-kesintisiz-servis) (background process).

For example, while the main agent writes code, a Sidecar agent might just watch for file changes in a folder (File Watcher) or monitor the system's RAM usage and feed live telemetry data back to the main [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model).

## Simple Reflex Agents

These are the most primitive type of autonomous software — they don't remember the past (no [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek)), don't [plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) for the future (no [Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) capability), and just produce an instant reflexive response to the current situation (input).

They work on logic like "if the [temperature](https://oguzerol.com/en/autonomous-ai-systems-glossary/#temperature-sicaklik) goes above 30, turn on the AC." In advanced [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems, these agents are used for low-cost [routing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#redirect-yonlendirme) tasks ([Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics)).

## Sitemap

This is the file that presents search engines with a list of every page on a website in XML format (usually found at \`/sitemap.xml\`).

It speeds up [Google's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) [crawling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#crawling-tarama)-based [discovery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) of your site and eases the [Indexing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#indexing-indeksleme) process. It's submitted through Google Search Console. Autonomous agents can spot missing pages or errors and update the sitemap.

## Skills / Agent Skills

These are standard-format ([SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd)) instruction manuals that explain "how the work should be done," turning an agent from a dry [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) into an enterprise-grade worker.

Where [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) provide the physical power to act, Skills are the recipe for what order to use those [tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) in, with what security standards, and with what architecture ([Procedural Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#procedural-memory-prosedurel-hafiza)).

## SKILL.md

This is the core skill document, written in YAML metadata (frontmatter) plus Markdown, that explains how agents should do a specific job (e.g., a database migration).

By giving the agent [Few-Shot Prompting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#few-shot-prompting) (example-based guidance), it guarantees the work gets done exactly in the format the developer wants ([Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti)) and follows [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) (security) rules.

## Slack Integration

This is the integration that lets autonomous agents send task notifications, error reports, and status updates to the Slack team-communication platform.

It's one of the channels of the [Notification System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#notification-system-bildirim-sistemi). It connects via Webhook or an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu). Autonomous agents can send an automatic alert to Slack when [Campaign Syncing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#campaign-syncing) finishes or when [Budget Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#budget-guardrails) are exceeded.

## Slash Commands

These are keyboard [shortcuts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) built to let users quickly trigger the autonomous agent ecosystem through the interface (e.g., \`/[plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)\`, \`/[execute](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula)\`).

Instead of writing out long [Prompt Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-engineering), they drop the agent straight into a specific [workflow](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar). Especially on [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) platforms, they're the most practical way to put agents into a specific mode (e.g., test-writing mode).

## Smart Voice Transcription

This is the technology that lets a developer communicate with agents by voice — like talking to a teammate — without using a keyboard, while the system instantly transcribes that voice with high accuracy, using technical terms appropriate to the context, and hands it to the agent.

It lets a human ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) quickly explain what they need during the [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) or [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) stages of long, complex projects.

## SOP (Standard Operating Procedure)

These are the strict, [repeatable](https://oguzerol.com/en/autonomous-ai-systems-glossary/#idempotent-tekrarlanabilir) work processes an enterprise company or project defines as "this is how we write code here" — the kind of thing an agent needs to know by heart.

[Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems eliminate human error by reading these SOPs through [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd) or [GEMINI.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#geminimd---agentsmd) files. It's the core key to getting deterministic, predictable output.

## Spark (MCP Host)

This is one of the open-source or proprietary [orchestration](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestration-orkestrasyon) hosts that easily manages and sets up [MCP (Model Context Protocol)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) servers in the [background](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) and puts them at an agent's ([Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#client-istemci---mcp)) service.

It keeps a developer from getting lost in [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) screens; it speeds up the connection infrastructure that lets an agent talk to the outside world through [Dynamic Tool Discovery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-tool-discovery).

## Sparse Persistence

This is a strategy where, instead of saving all the meaningless data and temporary logs it produces to permanent memory, an agent writes only the truly important decision moments, errors, and successful strategies sparsely to a database ([Mem0](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mem0-kalici-altyapi), [Firestore](https://oguzerol.com/en/autonomous-ai-systems-glossary/#firestore-firebase)).

It saves memory to help solve [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi) and [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) problems.

## Spreadsheet Automation

This is the process of automating repetitive data processing, reporting, and analysis tasks in [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) Sheets or [Microsoft](https://oguzerol.com/en/autonomous-ai-systems-glossary/#copilot-github-microsoft) Excel.

Autonomous agents can pull Google Analytics (GA4) or Google Ads data and generate automatic reports. They can connect to [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) Sheets through an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) and autonomously update campaign performance tables.

## SQL

This is the standard programming language (Structured Query Language) used to query, insert, and update data in relational databases (MySQL, PostgreSQL).

Autonomous agents can run database queries with [CLI Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-tools). Using SQL on [BigQuery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#bigquery), they can run deep analysis on raw Google Analytics (GA4) data. It's the structural alternative to [NoSQL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nosql) databases.

## SSL Certificate

This is the digital security document that encrypts a website's communication with a user and provides a secure connection (HTTPS), shown with a "lock" icon in the browser.

It's an SEO ranking factor; sites without SSL (HTTP) get flagged as "Not Secure" by [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google). It's a mandatory part of [Domain](https://oguzerol.com/en/autonomous-ai-systems-glossary/#domain-alan-adi) configuration.

## State Graph

This is the method of building agents' [workflows](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar), on platforms like [LangGraph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#langgraph---langchain), as a visual logic graph made of [Nodes](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nodes-dugumler) and connections ([Edges](https://oguzerol.com/en/autonomous-ai-systems-glossary/#edges-kenarlar)), instead of text blocks where steps proceed randomly.

It makes tracking which stage the system is in ([State Tracking](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-tracking-durum-takibi)) mathematical, and makes rolling back to the start after an error autonomous and controlled.

## State Management

This is when an agent (or a swarm of agents) securely saves and manages its current progress, variables, and results (State) while carrying out a long-running task.

If the State is lost, the agent suffers the [Goldfish Memory Problem](https://oguzerol.com/en/autonomous-ai-systems-glossary/#goldfish-memory-problem) and forgets what it was doing. In [State Graph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-graph-durum-grafigi) structures especially, it ensures data is passed on completely as it moves from one node to the next.

## State Tracking

This is the mechanism by which the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) monitors, through a dashboard or a [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) object, exactly which stage each of the dozens of [Worker Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) below it is at, and who's succeeding versus stuck in an infinite loop.

It's the observation stage that prevents chaos in large [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) operations, checking whether work has hit [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi).

## State Verification

This is when an agent, after carrying out an [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) or updating a database, fires off a second confirmation query — "Did this actually happen?" — to prove the system's current state.

Not trusting an agent that says "Done" through [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) is the foundation of autonomous system security. It's the logical validation step of the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model).

## state.vscdb

Especially in VSCode and [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide) derivatives, this is the temporary [SQLite](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sql) (or [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json))-based system file that holds the IDE's and agent's local [session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) state, open tabs, and interface settings.

If it gets corrupted, an agent's [Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek) context can be damaged, but it usually doesn't leave permanent damage on [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek).

## Static Context

This is the method of piling an entire project, or massive library documents, into the [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) right from the start — whether or not the agent will actually use them.

While this approach works on small projects, on large ones it slows the agent down ([Latency](https://oguzerol.com/en/autonomous-ai-systems-glossary/#latency-gecikme-suresi)) and causes it to get stuck on unnecessary data, triggering [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) syndrome. The fix is [Dynamic Context](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-context-dinamik-baglam) ([Lazy Loading](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lazy-loading-tembel-yukleme)) architectures.

## Stochastic Multi-Agent Consensus

This is the process where AI [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) carrying randomness and probability ([Stochasticity](https://oguzerol.com/en/autonomous-ai-systems-glossary/#stochasticity-rastlantisallik)) produce different decisions inside [Debate Loops](https://oguzerol.com/en/autonomous-ai-systems-glossary/#debate-loops-tartisma-odalari), argue with each other, and eventually reach a shared consensus on the correct answer (or code).

It lets a wrong decision a single [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) might make ([Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon)) get filtered and purified through the multi-agent system's autonomous critique process.

## Stochasticity

This is [LLMs'](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) nature of not answering the same question (the same input) with the exact same words every single time — being probabilistic and variable, depending on the [Temperature](https://oguzerol.com/en/autonomous-ai-systems-glossary/#temperature-sicaklik) setting of generation.

It's an advantage in creative work, but dangerous in coding tasks, like software, that need determinism. [JSON Validation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-validation-dogrulama) and [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) are used to help agents get past this randomness.

## Structured Output

This is when an agent produces output in a machine-readable, parseable format directly usable for [Function Calling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#function-calling) (e.g., [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json), XML), instead of returning plain text like it's writing a story.

With its structure guaranteed by a [Schema Validator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#schema-validator) ([Pydantic](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pydantic), etc.), this output is mandatory for agents to be able to trigger external systems and [APIs](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api).

## Subagent

In [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) architecture, this is a narrowly-authorized agent created or called — usually by the [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef) — to carry out a specific sub-task (e.g., only browsing the web, or only writing [CSS](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript)).

It keeps the main agent's memory ([Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi)) from filling up with unnecessary detail. When its job is done, it reports the result it got (e.g., a [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) recording a video) back up and shuts down.

## Swarms

This is a [Decentralized Agent Networks](https://oguzerol.com/en/autonomous-ai-systems-glossary/#decentralized-agent-networks) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) where large numbers of lightweight, fast AI agents — usually built as [Reactive Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reactive-agents-reaktif-ajanlar) — come together with simple rules ([A2A](https://oguzerol.com/en/autonomous-ai-systems-glossary/#a2a-agent-to-agent)), like a swarm of bees or ants in nature, and solve a complex problem without a central dictator ([Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan)).

## System Prompt

This is the **"master system instruction"** that defines an agent's identity, role, permissions, and the core constitution it must follow.

Unlike the instant chat messages a user types, it's the main framework the agent must stay faithful to for its entire working life.

It should be protected with [Guardrails](https://oguzerol.com/en/autonomous-ai-systems-glossary/#guardrails-korkuluklar) against malicious [Prompt Injection](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-injection-zehirli-girdi) attempts.

## System Thinker

Unlike visionless agents that only write the small piece of code in front of them ([Doer](https://oguzerol.com/en/autonomous-ai-systems-glossary/#doer---executor-uygulayici)/[Executor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#doer---executor-uygulayici)), this is a strategically-thinking agent profile (usually a [Conductor](https://oguzerol.com/en/autonomous-ai-systems-glossary/#conductor-ana-sef) or [Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)) that has command of the whole project, its architecture, performance metrics, and [Clean Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#clean-architecture) rules.

## Task Artifact

This is a visual document ([Artifact](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar)) agents keep and present to a human, in a structured format (Markdown, table), recording which tasks are done, which are in progress (a checklist), and which decisions have been made (status updates) — so they don't get lost during long-running work or projects.

It makes it easier for us to [observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) the agent's progress.

## Task Plan (task.md)

This is the reference (to-do) document a [Planner Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#planner-agent-planlamaci-ajan) creates to reach the main goal, laying out step by step which [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) to use and which [Subagent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan) to assign.

It's the autonomous system's route. Agents keep updating this file throughout the loop to check whether the system has reached [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi).

## Technical Debt

This is a buildup of bad (spaghetti) code that violates [Clean Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#clean-architecture) rules in favor of a fast fix (a patch), making the project unmanageable down the line.

[Pragmatist agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#pragmatist-pragmatist-ajan) knowingly take on this debt to move faster, while [Reviewer Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) constantly work to refactor the code to pay it down, keeping the system from collapsing.

## Telegram Bot

This is an automated bot integration used to manage autonomous agents, send commands, and receive notifications through the Telegram messaging platform.

It's one of the [Notification System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#notification-system-bildirim-sistemi) channels. Autonomous agents can send campaign reports to Telegram at intervals set by [Cron](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cron-zamanlanmis-gorevler), or get [HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop) approval from a user.

## Temperature

This is the mathematical hyperparameter that determines how much risk (creativity) an [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) takes when choosing a word — closer to zero produces more deterministic, precise results (lower [Stochasticity](https://oguzerol.com/en/autonomous-ai-systems-glossary/#stochasticity-rastlantisallik)), closer to one produces more creative, random ones.

In autonomous agents' code-writing tasks, it's usually kept at 0 or very low to minimize [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) risk.

## Terminal

This is the command-line screen ([CLI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cli-komut-satiri-arayuzu)) where agents send commands directly to servers or a computer and get results back as logs, without using a graphical interface (GUI).

By giving agents direct terminal permission inside an [Agentic IDE](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ide), they can autonomously carry out physical [actions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) ([ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem)) like creating folders, compiling, and starting servers within [nsjail](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nsjail) ([Sandbox](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execution-environment-sandbox)) [boundaries](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar).

## Token

This is the basic unit a [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) ([LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin)) breaks the text (or code) it reads and produces into for meaning, used to measure billing and capacity (1 Token is usually roughly 3/4 of a word).

In agentic systems, [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) (memory capacity) and [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) (cost/budget) calculations are done entirely in this currency.

## Token Burn Rate

This is the amount of [Tokens](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b) (money) an [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) system — especially multi-agent setups stuck in a loop — burns through per second or minute while operating through an [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api).

Without being reined in by [Cost Control Limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#cost-control-limits), an agent stuck in an infinite loop can burn through thousands of dollars in [tokens](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b) by morning. Strategies like [Fast Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fast-mode-heuristics) and [Prompt Caching](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompt-caching-onbellekleme) are used to reduce this fuel consumption.

## Tool Operator

In hierarchical systems, this is a narrowly-authorized ([Narrow AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#narrow-ai-dar-yapay-zeka)) agent that has no direct thinking ([Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme)) or code-writing ability; its only job is to run one specific tool (e.g., only opening a browser, or only triggering an [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api)) on an order from the [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan).

It works like a [Subagent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan). It lightens complex systems and enforces separation of duties ([Principle of Least Privilege](https://oguzerol.com/en/autonomous-ai-systems-glossary/#principle-of-least-privilege)).

## Tools

These are the functions that keep an [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) ([model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model)) from being trapped in the digital world, letting it physically intervene in the outside world (a database, the [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal), the internet).

They're triggered during the [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage of the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) process. The [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) just decides "delete this file" (Reason) — [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-mcp-bileseni) are what actually carry out that [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem). It's the most fundamental weapon that sets agents apart from classic chatbots ([ChatGPT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chatgpt)).

## Tools (MCP Component)

This is the action-oriented one among the 3 core components an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) offers to an agent ([Client](https://oguzerol.com/en/autonomous-ai-systems-glossary/#client-istemci---mcp)) — the others being [Resources](https://oguzerol.com/en/autonomous-ai-systems-glossary/#resources-mcp-bileseni) and [Prompts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#prompts-mcp-bileseni).

It gives the agent capabilities specific to that server (e.g., a "Salesforce_Add_Customer" tool). The moment the agent connects to the server, thanks to [Dynamic Tool Discovery](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-tool-discovery), it learns this tool along with its [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) schema ([Self-Describing API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-describing-api)) and starts using it.

## Top-P

This is a mathematical hyperparameter, working together with [Temperature](https://oguzerol.com/en/autonomous-ai-systems-glossary/#temperature-sicaklik), that narrows the probability pool an [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) draws from when choosing a word.

It's kept low in autonomous agents' coding tasks to reduce [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) risk and get stricter, more predictable (deterministic) [Structured Output](https://oguzerol.com/en/autonomous-ai-systems-glossary/#structured-output-yapisal-cikti).

## Trajectory

This is the full route (footprint) and filmstrip of exactly which steps an agent took from start to finish to complete a complex task, which [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) it used in what order, and how it recovered from errors ([Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma)).

It captures not just the result but the process. [Episodic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#episodic-memory-epizodik-bellek) is built directly out of storing these Trajectory records.

## Transformer

This is the neural network architecture, invented by [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) in 2017, that forms the foundation of every modern [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) today ([OpenAI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#openai) GPT, [Claude](https://oguzerol.com/en/autonomous-ai-systems-glossary/#claude-anthropic), [Gemini](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google), [DeepSeek](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model)).

It uses an "Attention" mechanism to understand text, grasping how the words in a sentence relate to each other. It's the source of [Embeddings](https://oguzerol.com/en/autonomous-ai-systems-glossary/#embeddings-yerlestirmeler) and [Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reasoning-akil-yurutme) abilities.

## Tree of Thoughts (ToT)

This is a [Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) architecture used by advanced agentic systems ([Deep Reasoning](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deep-reasoning-derin-dusunme) [models](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), etc.) that goes beyond [Chain of Thought](https://oguzerol.com/en/autonomous-ai-systems-glossary/#chain-of-thought-cot) — building multiple probability branches (a tree) at once while solving a problem, evaluating those branches in parallel, and pruning the ones that hit a dead end.

It enables error-free decision-making on mathematical and complex software-architecture tasks.

## TUI (Terminal User Interface)

This is the graphical interface agents present in the [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) or command line, instead of a web browser (GUI).

Developers often prefer the TUI environment to manage an agent directly with the keyboard (e.g., [Slash Commands](https://oguzerol.com/en/autonomous-ai-systems-glossary/#slash-commands-egik-cizgi-komutu)), read [Walkthrough](https://oguzerol.com/en/autonomous-ai-systems-glossary/#walkthrough-yurutme-raporu) documents, and give [Interactive Approvals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#interactive-approvals), instead of using heavy web panels. It provides speed and simplicity.

## User Memory

This is a [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) unit where an agent stores not just the current project, but "the user's personal preferences" directly (e.g., "Oğuzhan likes dark mode, and always wants me to use Tailwind in code").

It's managed with systems like [Mem0](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mem0-kalici-altyapi). It lets the agent offer a personalized, user-specific experience in every new [session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum).

## User Override

This is when a human steps in ([HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop)) and directly overrides a decision or piece of code the autonomous agent produced inside the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu), saying "No, do it this way instead" — intervening directly in the process.

The agent detects this intervention during the [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage, changes course, and records it in [Reflective Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reflective-memory-yansitici-bellek) so it doesn't repeat the same mistake.

## Utility-Based Optimization

This is the logic where an agent, when deciding on an [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem), doesn't just follow the rules — it mathematically picks the most profitable option by weighing the decision's cost ([Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi)), speed ([Latency](https://oguzerol.com/en/autonomous-ai-systems-glossary/#latency-gecikme-suresi)), and success rate (benefit/cost).

It's used often, especially by finance and ad-optimization agents (ROAS).

## UI Testing

This is when agents like a [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) autonomously test the interface of a website they've written or changed (buttons, menus, colors) down to the pixel, using [Vision](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme) ability.

It doesn't just check whether the code compiles — it checks whether the screen a human sees is actually designed correctly.

## Vector Database

Unlike classic [SQL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sql) databases, this is a next-generation database type where text and code are stored as mathematical numbers with hundreds of dimensions ([Embeddings](https://oguzerol.com/en/autonomous-ai-systems-glossary/#embeddings-yerlestirmeler)) — e.g., Pinecone, Chroma.

It makes it possible to search by "meaning" rather than by letters ([Semantic Search](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-search-anlamsal-arama)). It's the heart of [RAG (Retrieval-Augmented Generation)](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rag-retrieval-augmented-generation) architectures.

## Verify

This is a logical check-up process inside the [Agentic Loop's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage. When an agent writes code, this is where it [verifies](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-validation-dogrulama) not just that the code "runs," but that it actually passes the tests in the [Definition of Done](https://oguzerol.com/en/autonomous-ai-systems-glossary/#definition-of-done-bitti-tanimi) criteria.

In the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model), the [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) does this [verification](https://oguzerol.com/en/autonomous-ai-systems-glossary/#json-validation-dogrulama).

## Vibe Coding

This is the practice of producing random code purely by chatting with AI, without writing any software architecture, rule set, or rigorous tests ([Evals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#evals-evaluation---hard-evals)).

While fun for quick prototyping, it creates fragile, unmaintainable systems in enterprise production environments.

The antidote is the [Agentic Engineering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-engineering) discipline, with its rigorous engineering rules.

## Video-to-Action Pipeline

This is the process where an agent like a [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) watches the screen (or a video stream) with a [Vision](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) before acting on a website, understanding button and menu positions in seconds and autonomously deciding what to click and which forms to fill in.

It removes the dependency on [DOM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dom-document-object-model) ([HTML](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript)) code, letting the agent navigate the site by actually seeing it, like a human.

## Vision

This is an [LLM's](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) ability to go beyond just reading text and actually see and analyze photos, screenshots, or designs it's given, like a human would.

It's the foundation of autonomous tasks that need visual intelligence, like [UI Testing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ui-testing-arayuz-testi) and [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) navigation.

## Walkthrough

This is the structured ([Artifact](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar)) report an agent presents to the end user (a human) after finishing a big task, summarizing step by step what it changed, which errors it fixed, and how the new architecture works.

It's written using [Progressive Disclosure](https://oguzerol.com/en/autonomous-ai-systems-glossary/#progressive-disclosure) logic. It lets a human grasp the entire [Trajectory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#trajectory-yurutme-gecmisi) without reading the code line by line.

## Web Scraper Tool

This is a dedicated external tool that reads the raw [HTML](https://oguzerol.com/en/autonomous-ai-systems-glossary/#css---html---javascript) from live websites and runs it through a [Curate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#curate---curation-kurasyon) process, turning it into clean Markdown or [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json) an [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) can read.

It's the biggest weapon [Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan) units have for gathering fresh data (e.g., competitor prices) during [Agentic RAG](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-rag) workflows.

## WhatsApp Business API

This is the official programming interface that lets businesses automatically message customers, send notifications, and provide support over WhatsApp.

It's used as a [Notification System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#notification-system-bildirim-sistemi) and lead-generation channel. Autonomous agents can autonomously send order confirmations, appointment reminders, and similar messages through Webhook triggers.

## Worker Agent

Inside a [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru), this is a specialist agent that takes orders from the top-level [Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan) or [Planner](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) agent and actually carries out a specific task (e.g., only writing code, only pulling data).

It takes on the "[Maker](https://oguzerol.com/en/autonomous-ai-systems-glossary/#zapier---make)" role in the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model). When its job is done, it presents a [Walkthrough](https://oguzerol.com/en/autonomous-ai-systems-glossary/#walkthrough-yurutme-raporu) and wraps up its task.

## Workflows

This is when agents autonomously carry out a complex task through precise, pre-designed steps ([Nodes](https://oguzerol.com/en/autonomous-ai-systems-glossary/#nodes-dugumler) and [Edges](https://oguzerol.com/en/autonomous-ai-systems-glossary/#edges-kenarlar)) and a set route ([State Graph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-graph-durum-grafigi)) on systems like [LangGraph](https://oguzerol.com/en/autonomous-ai-systems-glossary/#langgraph---langchain), instead of random loops.

It prevents infinite-loop spirals and ensures deterministic results that meet [Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim) standards.

## Working Memory

This is the RAM-like memory space, [short-term](https://oguzerol.com/en/autonomous-ai-systems-glossary/#short-term-memory-calisma-bellegi), that an agent ([model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model)) is actively working with in the currently open [session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) (or [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal) tabs).

This memory evaporates the moment the [session](https://oguzerol.com/en/autonomous-ai-systems-glossary/#session-oturum) closes. It's subject to [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) [limits](https://oguzerol.com/en/autonomous-ai-systems-glossary/#privilege-boundaries-sinirlar). Important data the agent needs to remember later must be moved from here into [Long-Term Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#long-term-memory-kalici-bellek) through [Memory Consolidation](https://oguzerol.com/en/autonomous-ai-systems-glossary/#memory-consolidation).

## Workspace Rules

These are local instructions — found only at the root of the specific project being worked on (e.g., in \`.agents\` or \`[GEMINI.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#geminimd---agentsmd)\`) — holding strict rules like "architecture, style, permissions" specific to that project.

The agent reads these rules first during the [Discover](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) stage. They take priority over [Global Skills](https://oguzerol.com/en/autonomous-ai-systems-glossary/#global-skills-kuresel-beceriler) (general-purpose capabilities) and prevent cross-project contamination.

## YAML Frontmatter

This is the machine-friendly header format, usually found at the top of skill files like [SKILL.md](https://oguzerol.com/en/autonomous-ai-systems-glossary/#skillmd), written between two \`---\` lines, that instantly gives the agent metadata like "What is this skill called, what does it do?"

Without needing to read the entire text, an agent can make [Semantic Routing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-routing-anlamsal-yonl) decisions in milliseconds using this [JSON](https://oguzerol.com/en/autonomous-ai-systems-glossary/#csv---json)/YAML data.

## Zapier / Make

These are automation platforms that connect different [applications](https://oguzerol.com/en/autonomous-ai-systems-glossary/#execute-uygula) (CRM, email, [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google) Sheets, social media) without code to build automated [workflows](https://oguzerol.com/en/autonomous-ai-systems-glossary/#workflows-is-akislari---sablonlar).

They work through Webhook triggers. They're the no-code alternative to the integrations autonomous agents can do directly through an [MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu) or [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api). [n8n](https://oguzerol.com/en/autonomous-ai-systems-glossary/#n8n) is the open-source rival.

## Zero-Shot Prompting

This is when an agent is given a task with a single, bare command — "do this" — with no past reference, training, or example ([Few-Shot Prompting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#few-shot-prompting)) provided at all.

The [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model) acts purely by relying on its own internal [Semantic Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#semantic-memory-anlamsal-bellek). It's not usually preferred in complex [Agentic AI](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-ai) systems because it carries a high risk of [Hallucination](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hallucination-halusinasyon) and wrong formatting.

## Iceberg Technique

This is a UX/UI technique where the information or [Artifacts](https://oguzerol.com/en/autonomous-ai-systems-glossary/#artifacts-artefaktlar) interface presented to agents shows only the small piece the user or agent needs "right now" above the waterline, just like an iceberg.

It's the foundation of [Progressive Disclosure](https://oguzerol.com/en/autonomous-ai-systems-glossary/#progressive-disclosure) logic. The deep technical detail underneath (the bottom of the iceberg) only surfaces when the agent hits an error or a human asks for detail — saving [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) space.

## Idempotent

This is the property where an operation an agent carries out during the [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage (e.g., clicking a button, writing a database record) doesn't create a new problem or extra data even if the same command runs thousands of times in a row — it gives the same result as the first run.

It's a reliable [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api) design rule that stops an agent stuck in an infinite loop from leaving behind damage that crashes the system, at the hardware level.

## Identity Propagation

In [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru) and [Multi-Tenant](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-tenant-coklu-kiraci-izolasyon) architectures, this is when a main agent (e.g., for Customer A) hands off a task to a [Subagent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan) ([Worker Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#worker-agent-isci-ajan)), carrying that customer's "[access](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reach-erisim) permission / [token](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-jeton---yapay-zeka-para-b)" info (identity) along with it.

It enforces the [Principle of Least Privilege](https://oguzerol.com/en/autonomous-ai-systems-glossary/#principle-of-least-privilege). A [Subagent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan) can never use more identity than the parent agent has.

## Implementation Plan

This is the engineering (implementation) contract prepared during the [Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla) stage, usually by [Deliberative / Planning Agents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deliberative---planning-agents), listing out all the architectural detail, the [Tools](https://oguzerol.com/en/autonomous-ai-systems-glossary/#tools-araclar---modelin-elleri) to be used, and the [SDLC](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sdlc-yazilim-yasam-dongusu) steps before coding even begins.

Once the agent finishes this [plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla), it usually requests approval from a human via [HITL](https://oguzerol.com/en/autonomous-ai-systems-glossary/#hitl-human-in-the-loop) ([Interactive Approvals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#interactive-approvals)) and moves to [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) only after approval.

## Implementer Agent

Inside a hierarchical [Multi-Agent System](https://oguzerol.com/en/autonomous-ai-systems-glossary/#multi-agent-system-mas---suru), this is a high-muscle specialist agent that doesn't build strategy — it writes the raw code, compiles the system, and triggers [APIs](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api), sticking strictly to the [Implementation Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#implementation-plan) document prepared by the [Planner Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#planner-agent-planlamaci-ajan).

It makes up the "[Maker](https://oguzerol.com/en/autonomous-ai-systems-glossary/#zapier---make)" (producer) [half](https://oguzerol.com/en/autonomous-ai-systems-glossary/#knowledge-items-kis) of the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) [model](https://oguzerol.com/en/autonomous-ai-systems-glossary/#deepseek-model). When it finishes its work, it submits its output for [Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#critic---reviewer-agent) review.

## Indexing

This is the process where [Google](https://oguzerol.com/en/autonomous-ai-systems-glossary/#gemini-google), after [discovering](https://oguzerol.com/en/autonomous-ai-systems-glossary/#discover-kesfet) a web page through [Crawling](https://oguzerol.com/en/autonomous-ai-systems-glossary/#crawling-tarama), analyzes its content and saves it into the search database.

Pages that aren't indexed never show up in search results (SERP). Indexing status can be checked through Google Search Console. [Robots.txt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#robotstxt) and [Sitemap](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sitemap-site-haritasi) files govern this process.

## Infinite Loop Mitigation

These are the hardware and software safety mechanisms that stop an autonomous agent from falling into an endless, unproductive loop by stubbornly retrying the same bug it can't solve, burning through system resources ([Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi)).

They include measures like a "maximum 5 tries" limit ([Rate Limiting](https://oguzerol.com/en/autonomous-ai-systems-glossary/#rate-limiting-hiz-kisitlamasi)) or "call a human if you loop" ([Fallback Strategy](https://oguzerol.com/en/autonomous-ai-systems-glossary/#fallback-strategy-yedek-plan)).

## Interactive Approvals

This is when the system, running in a safe mode ([Iterative Mode](https://oguzerol.com/en/autonomous-ai-systems-glossary/#iterative-mode-onayli-mod)), freezes an [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) before carrying out something dangerous during the [ACT](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) stage (e.g., deleting a database or making a payment) and waits for an instant "Y/N" (yes/no) approval from a human in the [terminal](https://oguzerol.com/en/autonomous-ai-systems-glossary/#terminal).

It's the interface equivalent of the [Request Review](https://oguzerol.com/en/autonomous-ai-systems-glossary/#request-review) process in autonomous systems.

## Iterate

This is when an agent, during the final [Observe](https://oguzerol.com/en/autonomous-ai-systems-glossary/#observe-gozlemle) stage of the [Agentic Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#agentic-loop-ajan-dongusu) process, hits a failure or new data and — instead of ending the loop — updates its route ([Trajectory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#trajectory-yurutme-gecmisi)), replans ([Plan](https://oguzerol.com/en/autonomous-ai-systems-glossary/#plan-planla)), and moves back to [action](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem).

It's the heart of the [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma) mechanism — the agent repeats the process until it succeeds (or hits a limit).

## Iterative Mode

Unlike [Always Proceed](https://oguzerol.com/en/autonomous-ai-systems-glossary/#always-proceed) mode, where the agent runs entirely on its own end to end, this is a safe collaboration mode where the agent pauses at every critical stage (planning, coding, testing) to present the process to a human, working with them step by step ([iteratively](https://oguzerol.com/en/autonomous-ai-systems-glossary/#iterate-yinele)).

It's mandatory in enterprise projects and [SDLC](https://oguzerol.com/en/autonomous-ai-systems-glossary/#sdlc-yazilim-yasam-dongusu) processes so a human doesn't lose control ([Governance](https://oguzerol.com/en/autonomous-ai-systems-glossary/#governance-yonetisim)).

## Evaluator-Optimizer Pattern

This is an autonomous quality-loop architecture where one agent produces content or code, and a second, auditing agent checks the quality and sends feedback to get the first agent to fix it.

It's built on the [Maker-Checker Architecture](https://oguzerol.com/en/autonomous-ai-systems-glossary/#maker-checker-architecture) principle; when a single [LLM](https://oguzerol.com/en/autonomous-ai-systems-glossary/#llm-model---beyin) call can't give a flawless answer in one shot, it perfects the output through [Self-Correction](https://oguzerol.com/en/autonomous-ai-systems-glossary/#self-correction-kendi-kendini-onarma).

This pattern is especially preferred for hitting [Evals](https://oguzerol.com/en/autonomous-ai-systems-glossary/#evals-evaluation---hard-evals) standards and driving errors down to zero in complex systems.

## Subagent-Driven Development

This is a software development methodology where a complex project gets broken up by the main conductor ([Orchestrator](https://oguzerol.com/en/autonomous-ai-systems-glossary/#orchestrator-sef-ajan)) into small, independent, focused [Subagents](https://oguzerol.com/en/autonomous-ai-systems-glossary/#subagent-alt-ajan).

Because each subagent has its own dedicated [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) space, it keeps the main agent's desk from filling up ([Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi)).

When a task finishes, the subagent's output is checked ([Verify](https://oguzerol.com/en/autonomous-ai-systems-glossary/#verify-dogrula)) and merged into the main system.

## Context Compaction

This is the process of autonomously summarizing past steps, decision history, and tool output into memory during long-running agent sessions, so the [Context Window](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-window-baglam-penceresi) doesn't fill up.

It solves an agent's [Lost in the Middle](https://oguzerol.com/en/autonomous-ai-systems-glossary/#lost-in-the-middle-orta-kaybi) problem and meaningfully cuts [Token Burn Rate](https://oguzerol.com/en/autonomous-ai-systems-glossary/#token-burn-rate-jeton-tuketim-hizi) costs.

It plays a critical role in [Working Memory](https://oguzerol.com/en/autonomous-ai-systems-glossary/#working-memory-calisan-bellek) management.

## Computer Use / GUI Agent

This is an autonomous interface architecture where an AI agent manages desktop applications not just through text commands or an [API](https://oguzerol.com/en/autonomous-ai-systems-glossary/#api), but by reading the screen and using mouse clicks and keyboard input, like a human.

It goes beyond [Browser Sub-Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#browser-sub-agent--browser) abilities, carrying out [actions](https://oguzerol.com/en/autonomous-ai-systems-glossary/#act-eylem) at the entire operating-system level.

It analyzes the screen with a [Vision](https://oguzerol.com/en/autonomous-ai-systems-glossary/#vision-goruntu-isleme---gorme) model and is used in [UI Testing](https://oguzerol.com/en/autonomous-ai-systems-glossary/#ui-testing-arayuz-testi) or automating legacy systems.

## Tool Calling vs MCP

This is the core mindset difference between two architectural approaches to giving AI function capability.

Traditional Tool Calling calls functions statically written into the codebase, while [MCP](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-model-context-protocol) dynamically [discovers](https://oguzerol.com/en/autonomous-ai-systems-glossary/#dynamic-tool-discovery) tools through a standard plug-and-play server ([MCP Server](https://oguzerol.com/en/autonomous-ai-systems-glossary/#mcp-server-mcp-sunucusu)).

MCP offers flexibility and security, while static Tool Calling is preferred for lighter, local operations.

## Agent Drift

This is when an agent, over long, multi-step tasks, gradually strays from its original main goal and [System Prompt](https://oguzerol.com/en/autonomous-ai-systems-glossary/#system-prompt-sistem-istemi) boundaries over time, due to the effects of [Context Rot](https://oguzerol.com/en/autonomous-ai-systems-glossary/#context-rot-baglam-curumesi).

This can lead the agent into logical mistakes or an [Infinite Loop](https://oguzerol.com/en/autonomous-ai-systems-glossary/#infinite-loop-mitigation).

Regular status checks ([State Verification](https://oguzerol.com/en/autonomous-ai-systems-glossary/#state-verification-durum-dogrulama)) and auditing agents ([Reviewer Agent](https://oguzerol.com/en/autonomous-ai-systems-glossary/#reviewer-agent-denetleyici-ajan)) are used to prevent agent drift.

## Related Concepts

For every concept defined in this glossary, I’ve written a separate, deeper page — definition, comparison table, real examples, and observations from my own work. You’ll find them all below, sorted by category.

**Core Concepts**

- [What Is an AI Agent](https://oguzerol.com/en/what-is-an-ai-agent/)
- [What Is Agentic AI](https://oguzerol.com/en/what-is-agentic-ai-a-plain-language-definition/)
- [Types of AI Agents](https://oguzerol.com/en/types-of-ai-agents/)
- [What Is a Multi-Agent System](https://oguzerol.com/en/what-is-a-multi-agent-system/)
- [What Is Hallucination: Why AI Makes Things Up](https://oguzerol.com/en/what-is-hallucination/)
- [What Is a Token: The Real Cost of Using AI](https://oguzerol.com/en/what-is-a-token-how-ai-usage-cost-is-calculated/)

**Comparisons**

- [AI Agent vs Chatbot](https://oguzerol.com/en/ai-agent-vs-chatbot/)
- [AI Assistant vs AI Agent](https://oguzerol.com/en/ai-assistant-vs-agent/)
- [AI Agent vs Automation (RPA)](https://oguzerol.com/en/ai-agent-vs-automation-rpa/)
- [AI Agent vs LLM](https://oguzerol.com/en/ai-agent-vs-llm/)
- [AI Agent vs Workflow](https://oguzerol.com/en/ai-agent-vs-workflow/)
- [MCP Server vs Zapier](https://oguzerol.com/en/mcp-server-vs-zapier/)
- [Prompt vs Skill vs Agent](https://oguzerol.com/en/prompt-vs-skill-vs-agent/)
- [Claude vs ChatGPT](https://oguzerol.com/en/claude-or-chatgpt/)
- [No-Code vs Vibe Coding](https://oguzerol.com/en/no-code-vs-vibe-coding/)

**MCP, Claude Skills and Tools**

- [MCP and API Guide for Digital Marketers](https://oguzerol.com/en/mcp-and-api-guide-for-digital-marketers-2026/)
- [What Are Claude Skills, What Do They Do](https://oguzerol.com/en/what-is-claude-skills-what-is-it-for/)
- [How to Write a SKILL.md](https://oguzerol.com/en/how-to-write-skill-md/)
- [Is It Safe to Give an AI Agent Account Access](https://oguzerol.com/en/is-it-safe-to-give-an-ai-agent-account-access/)
- [What Is Claude AI, What Does It Cost](https://oguzerol.com/en/what-is-claude-ai-what-does-it-cost-how-to-use-it-from-turkey/)
- [What Is Vibe Coding](https://oguzerol.com/en/what-is-vibe-coding-what-a-non-coder-can-and-cant-do/)

**Automation**

- [What Is AI Automation — For Small Business](https://oguzerol.com/en/what-is-ai-automation-a-small-business-guide/)

[Back to top ↑](https://oguzerol.com/en/autonomous-ai-systems-glossary/#icerik)

## You may also find these useful

[Types of AI Agents: How Many Are There?How many types of AI agents are there: reflex, goal-based, multi-agent, tool-using, autonomous. Which fits which job — with a table and real examples.](https://oguzerol.com/en/types-of-ai-agents/)[What Is a Multi-Agent System: How Agents Work TogetherA multi-agent system splits a task across AI agents with roles. How it works, when it beats a single agent, and what it costs, based on Anthropic posts.](https://oguzerol.com/en/what-is-a-multi-agent-system/)[MCP Server vs. Zapier: A Guide for Non-CodersZapier is a pre-built automation bridge; MCP is a key handed to your AI assistant. See the difference, a comparison table, and how to choose.](https://oguzerol.com/en/mcp-server-vs-zapier/)

[Browse all articles →](https://oguzerol.com/en/articles/)