tm-ai 2.0.7

Time Machine for AI Agents — Cognitive Version Control for LLM context

🧠 CVC — Cognitive Version Control

Time Machine for AI Agents

Git for code. CVC for context.
Save. Branch. Rewind. Merge. — Your AI agent just got an undo button.

# 🎧 macOS / Linux / WSL
curl -fsSL https://jaimeena.com/cvc/install.sh | bash

# 🪟 Windows PowerShell
irm https://jaimeena.com/cvc/install.ps1 | iex

� Install via PyPI: pip install "tm-ai[all]" — Full docs at jaimeena.com/cvc →

📢 Latest Updates

• OpenClaw Integration (Beta): OpenClaw is now successfully installed on our Mac Mini host! We are currently working on a deep integration between CVC and OpenClaw to enable seamless cognitive versioning for all OpenClaw agents.
• Enhanced Agent Tools: v1.8.50 includes updated prompt instructions for the built-in agent to better handle high-volume token responses and complex JSON structures.

Why CVC?	Features	Quick Start	Docs	Architecture
Integrations	Community	Development	Roadmap	FAQ
� jaimeena.com/cvc — Full Docs & Install Guide

---

---

🚀 CVC 2.0: The Agentic Era

CVC has evolved from a simple context backup tool into a Cognitive Version Control system designed for Autonomous Agent Orchestration.

• Semantic Merging: Resolve cognitive collisions between branches using ChromaDB vector embeddings. Merges insights, not just logs.
• Quantum Branching (Tree of Thoughts): Race multiple LangGraph agents down parallel reality branches simultaneously. Merge the winning solution and archive failed attempts as learning data.
• Micro-Rollbacks: Rewind execution histories at the thought-level or tool-call level using the new THOUGHT_STEP and TOOL_CALL commit states.
• Context Distillation Checkpoints: Compress massive timelines dynamically to save LLM tokens while retaining semantic searchability.
• Multi-Agent Telepathy: Pack a branch's exact cognitive state into a portable .cvcpack and beam it over the MCP protocol to another machine.
• Claude-style Parallel Tool Execution: The CVC CLI Agent now supports running multiple destructive/write operations simultaneously via dynamic telepathy branching.

💥 The Problem

Your AI coding agent is brilliant — for about 20 minutes.

Then it forgets what it already fixed, contradicts its own plan,
and loops on the same error for eternity.

Sound familiar?

😵‍💫 Context Rot After 60% context fill, quality falls off a cliff

🔁 Error Loops Same mistake, different turn

🧠 No Memory Can't remember what it just did

💸 Token Waste Re-processing the same context

Bigger context windows don't fix this.

They just give the problem more room to spread.

✨ The Solution: CVC

Give AI agents what they've never had: memory that actually works.

---

🎯 Why CVC?

CVC is Git for the AI's brain.
Instead of versioning source code, it versions the agent's entire cognitive state — every thought, every decision, every conversation turn — as an immutable, cryptographic Merkle DAG.

💾 Save	🌿 Branch	⏪ Rewind	🔀 Merge	🔍 Search
Checkpoint the agent's brain at any stable moment	Explore risky ideas in isolation without polluting main context	Stuck in a loop? Time-travel back instantly	Merge learnings back — semantic, not syntactic	Find when the agent solved similar problems before

📊 Research-Backed Results

58.1% Context reduction via branching ContextBranch paper →

3.5× Success rate improvement with rollback GCC paper →

~90% Cost reduction through prompt caching Caching study →

~85% Latency reduction on restores Anthropic docs →

---

🚀 Features

🤖 Built-in Agent	🔌 Universal Proxy	⏱️ Time Machine
Just type cvc for a powerful AI coding assistant with 17 built-in tools and 5-provider support	Run any AI tool through CVC's time machine via API proxy or MCP server Zero configuration required	Auto-checkpoint every N turns Never lose context Crash recovery built-in Time-travel to any point --no-think for faster responses

🆚 How CVC Compares

FEATURE	CLAUDE CODE / CODEX	ANTIGRAVITY / CURSOR / VS CODE	🔥 CVC AGENT
Context Memory	Linear (Lost on restart)	Session-based / Linear	✅ Time-Travel (Project Isolated)
Branching	No branching / rollback	Not supported	✅ Full Branch Support
Undo Capability	Single step (maybe)	Single step / Ctrl+Z	✅ Instant Rewind (Any State)
Search History	Current session only	Current session only	✅ Global Semantic Search
Providers	Single provider (usually)	Vendor locked often	✅ Agnostic (5+ Providers)
Cost	Full Re-prompting	Full Re-prompting	✅ ~90% Cheaper (Cached)
Local / Offline	Cloud Dependent	Cloud Dependent	✅ 100% Local / Offline Capable
Context Merging	Not available	Not available	✅ Semantic Merge
Image Analysis	Limited	Limited	✅ Built-in Vision
Auto-checkpoint	Not available	Not available	✅ Configurable
Crash Recovery	Session lost	Session lost	✅ Full Restoration

---

⚡ Quick Start

Installation

🎯 Recommended: One-Line Installer (Installs Python if needed)

# 🎧 macOS / Linux / WSL
curl -fsSL https://jaimeena.com/cvc/install.sh | bash

# 🪟 Windows PowerShell
irm https://jaimeena.com/cvc/install.ps1 | iex

📦 Alternative: pip / uv

# pip (all providers)
pip install "tm-ai[all]"

# uv (faster, recommended)
uv tool install "tm-ai[all]"

# Specific provider only
pip install "tm-ai[anthropic]"   # Claude
pip install "tm-ai[openai]"      # GPT
pip install "tm-ai[google]"      # Gemini

Usage

# 🤖 Agent Mode - Just type 'cvc'
cvc

# 🔌 Proxy Mode - Zero-config launch
cvc launch claude          # Claude Code CLI
cvc launch aider           # Aider
cvc launch cursor          # Cursor IDE

# Or manual setup
cvc up                     # Setup + init + serve

Set API Keys

# Environment variables
export ANTHROPIC_API_KEY="sk-ant-..."
export OPENAI_API_KEY="sk-..."
export GOOGLE_API_KEY="AIza..."

# Or interactive setup
cvc setup

That''s it! 🎉

🌐 Full documentation, installation guides, and examples at jaimeena.com/cvc

---

📖 Documentation

🤖 Agent CLI Guide Complete guide to the built-in agent	⌨️ Slash Commands All slash commands reference	🔧 Tools Reference 17 built-in tools documentation	🔌 MCP Integration Model Context Protocol setup
🗺️ Cross-Mode Guide Agent + Proxy + MCP workflows	📁 Multi-Workspace Multiple project management	📚 Full Docs Complete documentation	📝 Changelog Version history
🌐 Official Website Full docs, installation guides, examples & more at jaimeena.com/cvc

---

🔌 Integrations

Run ANY AI tool through CVC''s Time Machine

Tool	How to Connect	Command
💎 Claude Code CLI	export ANTHROPIC_BASE_URL=http://127.0.0.1:8000	cvc launch claude
🛠️ Aider	Standard OpenAI-compatible endpoint	cvc launch aider
⌨️ Codex CLI	model_provider = "cvc" in config	cvc launch codex
🖱️ Cursor	Settings → Override Base URL	cvc launch cursor
💎 VS Code + Copilot	BYOK or MCP integration	cvc launch code
🏄 Windsurf	MCP integration	cvc launch windsurf
🚀 Antigravity	MCP integration	cvc mcp
🔄 Continue.dev	Base URL → http://127.0.0.1:8000/v1	cvc serve
🤖 Cline	Base URL → http://127.0.0.1:8000/v1	cvc serve
🦜 LangChain	Use CVC''s function-calling tools	cvc serve
👥 CrewAI	Use CVC''s function-calling tools	cvc serve
🤝 AutoGen	Use CVC''s function-calling tools	cvc serve
🌐 Open WebUI	Standard OpenAI-compatible endpoint	cvc serve

🔑 Auth pass-through: External tools can send their own API keys — CVC forwards them to the upstream provider.

Run cvc connect for interactive setup instructions.

---

🗝️ Architecture

Three-Tiered Local Storage

🗄️ SQLite Commit graph Branch pointers Metadata Fast traversal, zero-config

📦 CAS Blobs Context snapshots Zstandard compression Content-addressable Deduplicated, efficient

🔍 Chroma Semantic embeddings Vector search Optional "Have I solved this before?"

✨ Everything stays in .cvc/ inside your project
🔒 No cloud • No telemetry • Your agent''s thoughts are yours

📦 Project Structure

cvc/
├── agent/          # Built-in AI coding agent
│   ├── chat.py     # REPL loop, slash commands
│   ├── llm.py      # Unified LLM client (4 providers)
│   ├── tools.py    # 17 tool definitions
│   └── executor.py # Tool execution engine
├── adapters/       # Provider-specific formatting
│   ├── anthropic.py # Prompt caching support
│   ├── openai.py
│   ├── google.py
│   └── ollama.py
├── core/           # Data layer
│   ├── models.py   # Merkle DAG, Pydantic schemas
│   └── database.py # SQLite + CAS + Chroma
├── operations/     # CVC engine
│   ├── engine.py   # Commit, branch, merge, restore
│   └── state_machine.py # LangGraph routing
└── vcs/            # Git bridge
    └── bridge.py   # Shadow branches, Git notes, hooks

---

🤖 Supported Providers

Provider	Default Model	Alternatives	Notes
🟣 Anthropic	claude-opus-4-6	claude-opus-4-5, claude-sonnet-4-5, claude-haiku-4-5	Prompt caching
🟢 OpenAI	gpt-5.2	gpt-5.2-codex, gpt-5-mini, gpt-4.1	Auto prefix caching
🔵 Google	gemini-3.1-pro-preview	gemini-3-pro-preview, gemini-3-flash-preview	Multimodal + reasoning
⚪ Ollama	qwen2.5-coder:7b	qwen3-coder:30b, devstral:24b, deepseek-r1:8b	100% local, no API key
🟤 LM Studio	qwen2.5-coder-32b-instruct	devstral-small-2505, deepseek-r1-distill-qwen-32b, gemma-3-27b-it	100% local, OpenAI-compat

---

💬 Community

Join the CVC community

🐛 Issues Report bugs & request features

💡 Discussions Ask questions & share ideas

🌐 Website Docs, guides & examples

---

🛠️ Development

Contributions welcome! See below for setup instructions.

All contributions go through code review via Pull Requests.

Fork → Branch → Commit → Pull Request

🛠️ Dev Setup

# Clone the repository
git clone https://github.com/mannuking/cvc.git
cd cvc
uv sync --extra dev

---

🗺️ Roadmap

Feature	Status
🤖 Built-in Agent CLI	✅ Shipped v1.6.2 — 17 tools, 5 providers, --no-think
☁️ All 5 Provider Adapters	✅ Shipped — Anthropic, OpenAI, Google, Ollama, LM Studio
🔌 MCP Server	✅ Shipped — stdio + SSE transports
🚀 Zero-config Launch	✅ Shipped — cvc launch for all tools
🔗 Git Bridge	✅ Shipped — Shadow branches, hooks, notes
🎨 VS Code Extension	📜 Q2 2026 — Visual commit graph, time-travel
🌐 Web UI	📜 Q2 2026 — Browser visualization & management
👥 Multi-agent support	📜 Q3 2026 — Shared DB with conflict resolution
☁️ Cloud sync	📜 Q3 2026 — S3/MinIO for teams
📊 Metrics dashboard	📜 Q4 2026 — Cache hits, context utilization, analytics

---

❓ FAQ

What is CVC?

CVC (Cognitive Version Control) is Git for AI agent context. It versions the agent''s entire cognitive state as an immutable Merkle DAG, enabling time-travel, branching, and merging for AI conversations.

How is CVC different from Claude Code or Cursor?

CVC adds: ⏪ Time-travel • 🌿 Branching • 🔍 Semantic search • 🔀 Context merging • 💾 Auto-checkpoint • 🔄 Crash recovery • 🤖 Multi-provider support (5 vs 1)

Does CVC work with my existing AI tools?

Yes! CVC works in two modes:

1. Agent mode — built-in AI assistant
2. Proxy mode — transparent proxy for Claude Code, Aider, Cursor, VS Code, Windsurf, Continue, Cline, LangChain, CrewAI, AutoGen, and more

Is my data sent to the cloud?

No. Everything stays local in .cvc/. CVC has:
🔒 No telemetry • 🔒 No cloud sync • 🔒 No data collection
Your agent''s thoughts are yours.

How much does it cost?

CVC is free and open-source (MIT license). You only pay for LLM API calls (Anthropic/OpenAI/Google), and CVC makes them ~90% cheaper via prompt caching and 58.1% less context via branching. Use Ollama for 100% free local operation.

Can I use CVC offline?

Yes! Use Ollama with local models like qwen2.5-coder:7b, deepseek-r1:8b, etc. 100% local, no API key needed.

What providers are supported?

• 🟣 Anthropic (Claude Opus 4.6, Sonnet 4.5, Haiku 4.5)
• 🟢 OpenAI (GPT-5.2, GPT-5.2-Codex, GPT-5-mini)
• 🔵 Google (Gemini 3 Pro/Flash Preview, Gemini 2.5 Pro/Flash — with thinking level control)
• ⚪ Ollama (Qwen, DeepSeek, Devstral, etc.)
• 🟤 LM Studio (any locally loaded model — OpenAI-compatible)

Is there a GUI?

Not yet! But coming soon:
🎨 VS Code extension (Q2 2026)
🌐 Web UI (Q2 2026)

For now, use the beautiful terminal UI with Rich formatting.

Can I use CVC in production?

Yes! CVC is production-ready (v1.8.50) with crash-resistant local storage, used by solo developers, teams, and enterprises.

---

📚 Research

CVC is grounded in published research

Paper	Key Finding
ContextBranch	58.1% context reduction via branching
GCC	11.7% → 40.7% success with rollback (3.5× improvement)
Merkle-CRDTs	Structural deduplication for content-addressable DAGs
Prompt Caching	Anthropic/OpenAI/Google token reuse patterns

---

📜 License

MIT License — See LICENSE for details.

Free and open-source. Contributions welcome!

---

✨ Because AI agents deserve an undo button ✨

Made with ❤️ by developers who got tired of AI agents forgetting what they just did.

---

Open Source — MIT License

---

_{Built with Python 🐍 • FastAPI ⚡ • SQLite 🗄️ • LangGraph 🦜 • Rich 🎨}

• 2026-03-24: Addressed rich text markdown editing in HydroMain's Venture Dashboard and proposed Project Aegis (native macOS sandbox for OpenClaw).