Documentation Index

---

Documentation Index

Getting Started

• Getting Started Guide — Install, env setup, and 3 levels (1 line → 3 lines → 5 lines)
• README — Installation and quick start
• High-Level API — generateText(), streamText(), generateImage(), generateVideo(), analyzeVideo(), generateMusic(), generateSFX(), performOCR(), agent(), and agency()
• Examples Cookbook — 17 worked examples and the runnable files under examples/, including QueryRouter host hooks and finalized agency streaming
• CHANGELOG — Version history and release notes

Architecture & Core Concepts

• Architecture Overview — Complete system architecture and design principles

Features & Capabilities

Planning & Orchestration

• Orchestration Guide — Graphs, workflows, missions, voice nodes, checkpointing, YAML authoring
• Unified Orchestration Layer — One runtime, three authoring APIs (AgentGraph, workflow(), mission())
• AgentGraph — Full graph builder with typed nodes, conditional edges, and subgraphs
• workflow() DSL — Deterministic DAG pipelines with branching and parallel joins
• mission() API — Goal-first orchestration driven by the PlanningEngine
• Checkpointing — Resume, fork, replay, and memory consistency semantics
• Planning Engine — Multi-step task planning and execution
• Human-in-the-Loop — Approval workflows and human oversight
• Agent Communication — Inter-agent messaging and coordination

Safety & Security

• Guardrails System — Two-phase guardrail dispatch, the six built-in packs (PII redaction, ML classifiers, topicality, code safety, grounding guard, content policy rewriter), and the folder-level file permissions that Wunderland adds on top of AgentOS
• Safety Primitives — Circuit breakers, cost guards, stuck detection, and tool execution guards

Memory & Storage

• Cognitive Memory — Personality-modulated memory with Ebbinghaus decay, Baddeley's working memory, spreading activation, HEXACO-driven encoding, and the consolidation loop
• RAG Memory Configuration — Vector storage and retrieval setup
• SQL Storage Quickstart — Database integration guide
• Client-Side Storage — Browser-based persistence
• Immutable Agents — Sealing lifecycle, toolset pinning, secret rotation, and soft-forget
• Provenance Guide — HashChain, ChainVerifier, BundleExporter, proof levels, external anchors
• Provenance & Immutability — Sealed storage policy, signed ledger, and anchoring

Text Processing & Search

• Text Processing & Search: the section's map, what loads in a browser, and the runnable examples
• Text Processing Pipeline: tokenizers for prose and code, stop words, Porter stemming and the three preset pipelines
• Text Statistics and Classification: StatisticalUtilityAI, LLMUtilityAI and HybridUtilityAI for sentences, n-grams, keywords, readability, similarity, language, summaries, sentiment and classification
• Lexical and Hybrid Search: LexicalIndex, BM25Index, FuzzyIndex, HybridSearcher and reciprocal rank fusion
• Chunking and Snippets: SemanticChunker, chunkTurns, ChunkingEngine and snippetAround
• Reranking: RerankerService and the local, Cohere and LLM-judge rerankers

AI & LLM

• Structured Output — JSON schema validation and structured generation
• Streaming Semantics — Raw live chunks vs finalized approved output across textStream, fullStream, text, and finalTextStream
• Evaluation Guide — Test cases, graders, LLM-as-judge, A/B testing, experiment tracking
• Evaluation Framework — Testing, scoring, and quality assurance
• GMI versus agent() Study: the study that compares a GMI, agent() and a bare model loop on companion conversation; its page sets out the design and lists each report set once it is published
• Query Router — Tiered query classification, retrieval routing, keyword fallback, and grounded answer generation
• Image Generation Guide — Provider-agnostic image generation across cloud and local backends
• Capability Discovery Guide — Three-tier semantic discovery, CAPABILITY.yaml, meta-tool
• Capability Discovery — Full architecture reference
• Cost Optimization — Token usage and API cost management
• Cache Diagnostics — Anthropic cache-miss root-causing: per-step cache_miss_reason via auto-threaded request comparison

Extensions & Customization

• RFC Extension Standards — Extension development guidelines
• Self-Extension: Forging and Self-Improvement — Forged tools, self-improvement tools and specialist spawning
• Skills (SKILL.md) — Prompt modules loaded from directories/registries

Channels & Social

• Channels Guide — Multi-channel adapters with setup for Discord, Slack, Telegram, Twitter, and WhatsApp
• Social Posting Guide — SocialPostManager, content adaptation, scheduling, analytics

Platform & Infrastructure

• Platform Support — Supported environments and requirements
• Observability (OpenTelemetry) — Tracing, metrics, and log correlation/export (opt-in)
• Logging (Pino + OpenTelemetry) — Structured logs, trace correlation, and OTEL LogRecord export (opt-in)
• API Key Rotation: comma-separated keys per provider as a pool: ordered or round-robin, failover when a key is revoked, unpaid or throttled, probes and persisted state
• CLI Error Codes: the codes a CLI subprocess bridge raises, and when the Claude Code and Gemini CLI providers raise each one

Ecosystem

• Ecosystem — Related packages, extensions, and resources
• Releasing — Automated release process

API Reference

• TypeDoc API — Auto-generated API documentation

---

Quick Links

Resource	Link
Website	agentos.sh
GitHub	framerslab/agentos
npm	@framers/agentos
Issues	GitHub Issues

---

How to Use This Documentation

1. New to AgentOS? Start with the README for installation and basic usage
2. Understanding the system? Read the Architecture Overview
3. Building features? Check the relevant feature guide (Planning, HITL, Guardrails, etc.)
4. API details? Browse the TypeDoc API Reference
5. Troubleshooting? See Platform Support

---