CVC — Cognitive Version Control

Python/uv · Full system architecture · Agent AI layer + Storage + Distribution

Developer CLI User ──────── Telegram User CVC CLI tm-ai / uv cvc <command> Telegram Bot python-telegram bot API VS Code Ext vscode-extension/ marketplace gateway + proxy Gateway cvc/gateway.py request router Proxy cvc/proxy.py Auth cvc/auth.py token / session Agent Core cvc/agent/ • llm.py • chat.py • tools.py • sessions.py LLM Providers GitHub Copilot claude-sonnet-4.6 NVIDIA NIM Nemotron 120B Google Gemini Pro / Flash Core Modules cvc/core/ snapshot.py Merkle DAG commits models.py Commit / Branch / Agent skill_graph.py skill DAG traversal learning_extractor.py auto-extract noemata pageindex.py semantic search index Persistence core/persistence.py SQLite + .cvc/ blobs Hive Mind agent/memory.py shared agent memory COGNOME Engram compiler L1 smart context Git Integration agent/git_integration .cvc/ ↔ git repo GitHub sync_push / pull Hooks agent/hooks Cost Tracker cost_tracker.py token $$$ Legend Frontend / Entry Backend / Agent Cloud / LLM Database / Storage Security / Auth Async / Optional

Entry Points

  • • CVC CLI via tm-ai / uv
  • • Telegram Bot (python-telegram-bot)
  • • VS Code Extension (marketplace)
  • • All route through Gateway + Proxy layer

Agent Intelligence

  • • Agent Core: llm, chat, tools, sessions
  • • COGNOME Layer 1 — Engram compiler
  • • Hive Mind — shared memory across agents
  • • Skill Graph — DAG-based skill routing
  • • Learning Extractor — auto noemata

Storage & Distribution

  • • SQLite + .cvc/ blob store
  • • Merkle DAG commit graph
  • • Git sync: push/pull to GitHub
  • • jaimeena.com/cvc — install endpoint
  • • PyPI via tm-ai package