How key1lee/ai-os works
AI-OS is a shared engineering layer for otherwise independent projects; its documented Py.Dev runtime provides a provider-neutral model router, while projects retain their own rules, permissions, deterministic checks, and state. The local caller submits JSON through the CLI, and the router applies shared configuration and policy before provider handling. The tree and excerpts do not show the CLI entry point or provider dispatch calls, so those links are omitted; provider adapters are shown as separate runtime components.
Main entry points: router.py Model Router · README.md Engineering Standards · openai.py OpenAI Provider · parsed.py Parsed Provider
Components 9
Shared AI Layer
- Engineering Standards
README.md - Reusable Skills
skills
Py.Dev Runtime
- Model Router
py_dev/router.py - Context Policy
py_dev/context_policy.py - Intelligence
py_dev/intelligence.py - OpenAI Provider
py_dev/providers/openai.py - Parsed Provider
py_dev/providers/parsed.py
Other components
- Independent Projects
- JSON Caller
Connections 4
- JSON Caller → Model Router: submits JSON
README.md - Model Router → OpenAI Provider: provider dispatch
- Model Router → Parsed Provider: provider dispatch
- Independent Projects → Engineering Standards: uses shared guidance
README.md