Skip to content

Cognitive OS v3.0 — Universal AI Agent Operating System

Choose a tag to compare

View all tags
@aegntic aegntic released this 06 Jul 15:51
· 35 commits to main since this release
v3.0.0
f6772d9

Cognitive OS v3.0

The future is not predicted. It is compiled.

Cognitive OS is a universal cognitive operating system specification for AI coding agents. It defines a production-grade, 16-section cognitive architecture that governs how LLM-powered agents think, plan, verify, and execute across real-world systems.

A single master specification (AGENTS.md, 699 lines) auto-adapts to 5 AI agent harnesses through thin adapters: Claude Code, Cursor, Codex CLI, Gemini CLI, and Factory Droid.

What's Included

Cognitive Architecture (16 Sections)

  • 10-layer cognitive stack (intake triage T0-T4, reasoning modes, context budgeting, diff-aware editing, tool blindness prevention, semantic search, generated file awareness, phased execution, rollback planning, dependency discipline)
  • Multi-model orchestration with task complexity routing matrix and ensemble verification
  • 4-layer agentic memory stack (working, episodic, semantic, procedural)
  • Autonomous verification system with mandatory 7-step verification chain and self-healing pipeline
  • Swarm coordination protocol with sub-agent swarming, worktree isolation, and DAG execution
  • Adversarial security framework with prompt injection defense and circuit breakers
  • Economic intelligence with token cost tracking and ROI-aware task routing
  • Self-improvement pipeline with continuous learning loop and instinct evolution
  • Real-world actuation protocols for browser automation, deployments, and trading systems
  • Observability and tracing with quality gates and eval-driven development
  • Code quality standards for TypeScript/React, Python, Rust, and Solidity/DeFi
  • SPARC methodology integration (Specification, Pseudocode, Architecture, Refinement, Completion)

Multi-Harness Adapters

Harness File Purpose
Factory Droid .factory/AGENTS.md MCP servers, sub-agent strategy, custom droids
Claude Code CLAUDE.md ECC agent registry, gstack, model routing
Cursor .cursorrules Compact subset for code completion
Codex CLI CODEX.md OpenAI-optimized directives
Gemini CLI GEMINI.md Google-optimized extensions

Infrastructure

  • CI/CD pipeline (7 jobs: lint, format, test, link-check, tech-debt-scan, flaky-test-detection, docs-validation)
  • CodeQL security analysis (runs on push, PR, and weekly schedule)
  • Stale issue/PR bot (30-day stale, 7-day close, security exemptions)
  • Automated release workflow
  • 110 vitest tests across 5 test suites
  • Pre-commit hooks (husky + lint-staged + markdownlint + prettier)
  • Branch protection (enforce admins, no force push)
  • Secret scanning with push protection
  • Dependabot weekly updates
  • Devcontainer configuration
  • Architecture diagrams (Mermaid) and operational runbooks

Quick Start

git clone https://github.com/aegntic/cognitive-os.git
cd cognitive-os
npm install
npm test

Copy AGENTS.md and your harness adapter into any project. Your AI agent auto-detects and loads the spec at session start.

Documentation

Links

Assets 2
Loading