Skip to main content

AgentOS

Autonomous AI agent framework with cognitive memory, graph-based RAG, and HEXACO personality modeling

$npm install @framers/agentos

Get Started API Reference ⭐ GitHub

Get Running in Minutes

Three lines to a streaming AI agent. Add guardrails, tools, and memory as you need them.

🚀 Quick Start
import { AgentOS } from '@framers/agentos';

const agent = new AgentOS();
await agent.initialize({
  llmProviders: [{
    providerId: 'openai',
    apiKey: process.env.OPENAI_API_KEY,
    models: [{ modelId: 'gpt-4o' }],
  }],
});

// Stream a chat response
for await (const chunk of agent.processRequest({
  textInput: 'What are the key principles of distributed systems?',
  userId: 'user-1',
  sessionId: 'session-1',
})) {
  if (chunk.type === 'TEXT_DELTA') process.stdout.write(chunk.textDelta);
  if (chunk.type === 'FINAL_RESPONSE') console.log('\n✓ Done');
}

What's Inside

Cognitive Memory

Observational memory with Ebbinghaus decay, Baddeley working memory, HyDE retrieval, and personality-driven encoding.

Deep Research

LLM-as-judge query classification, 3-phase research pipeline, real-time progress streaming via SSE.

Safety & Guardrails

5 extension packs — PII redaction, ML classifiers, topicality, code safety, grounding guard. Two-phase parallel execution.

RAG & Vector Search

HNSW vector store, graph-augmented re-ranking, hybrid retrieval, adaptive HyDE thresholding.

Capability Discovery

Semantic tiered discovery — 89% token reduction, graph re-ranking, agents self-discover tools mid-conversation.

Extensions & Skills

45+ extensions, 37 channel adapters, 40 curated skills, multi-provider TTS/STT, image generation.

Planning & Orchestration

Multi-step planning engine, human-in-the-loop approvals, adaptive execution runtime.

API Reference

Auto-generated TypeDoc reference — every class, interface, type, and function documented.

Ecosystem

Core Platform

@framers/agentos

Modular orchestration runtime — the engine behind everything

@framers/agentos-extensions

45+ official extensions (channels, tools, integrations)

@framers/agentos-extensions-registry

Curated manifest builder — createCuratedManifest()

@framers/agentos-skills-registry

40+ curated SKILL.md prompt modules

@framers/agentos-skills

Skills runtime loader (reads SKILL.md → agent prompt)

Guardrail Extensions

@framers/agentos-ext-pii-redaction

4-tier PII detection (regex → NLP → NER → LLM)

@framers/agentos-ext-ml-classifiers

Toxicity, injection, jailbreak via ONNX BERT

@framers/agentos-ext-topicality

Embedding topic enforcement + drift detection

@framers/agentos-ext-code-safety

25 OWASP regex rules for LLM-generated code

@framers/agentos-ext-grounding-guard

RAG hallucination detection via NLI entailment

Apps & Tools

Agent runtime CLI — start, chat, seal, deploy

docs.agentos.sh

Documentation site (you are here)

agentos-workbench

Developer cockpit — debug sessions, inspect memory