Skip to main content

Time Machine for AI Agents โ€” Cognitive Version Control for LLM context

Project description

CVC Logo

๐Ÿง  CVC โ€” Cognitive Version Control

Time Machine for AI Agents

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


# ๐ŸŽง macOS / Linux / WSL
curl -fsSL https://jaimeena.com/cvc/install.sh | bash

# ๐ŸชŸ Windows PowerShell
irm https://jaimeena.com/cvc/install.ps1 | iex

PyPI version PyPI downloads Python 3.11+ License: MIT Code Style


๏ฟฝ Install via PyPI: pip install "tm-ai[all]" โ€” Full docs at jaimeena.com/cvc โ†’


Why CVC? Features Quick Start Docs Architecture
Integrations Community Development Roadmap FAQ
๏ฟฝ jaimeena.com/cvc โ€” Full Docs & Install Guide




๐Ÿ’ฅ The Problem


Your AI coding agent is brilliant โ€” 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?


๐Ÿ˜ตโ€๐Ÿ’ซ

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

๐Ÿ”

Error Loops
Same mistake,
different turn

๐Ÿง 

No Memory
Can't remember
what it just did

๐Ÿ’ธ

Token Waste
Re-processing the
same context

Bigger context windows don't fix this.

They just give the problem more room to spread.


โœจ The Solution: CVC

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




๐ŸŽฏ Why CVC?

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


๐Ÿ’พ
Save
๐ŸŒฟ
Branch
โช
Rewind
๐Ÿ”€
Merge
๐Ÿ”
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 โ€” semantic, not syntactic Find when the agent solved similar problems before

๐Ÿ“Š Research-Backed Results

58.1%

Context reduction
via branching

ContextBranch paper โ†’

3.5ร—

Success rate improvement
with rollback

GCC paper โ†’

~90%

Cost reduction through
prompt caching

Caching study โ†’

~85%

Latency reduction
on restores

Anthropic docs โ†’



๐Ÿš€ Features

๐Ÿค– Built-in Agent ๐Ÿ”Œ Universal Proxy โฑ๏ธ 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

๐Ÿ†š How CVC Compares

FEATURE CLAUDE CODE / CODEX ANTIGRAVITY / CURSOR / VS CODE ๐Ÿ”ฅ CVC AGENT
Context Memory Linear (Lost on restart) Session-based / Linear โœ… Time-Travel (Project Isolated)
Branching No branching / rollback Not supported โœ… Full Branch Support
Undo Capability Single step (maybe) Single step / Ctrl+Z โœ… Instant Rewind (Any State)
Search History Current session only Current session only โœ… Global Semantic Search
Providers Single provider (usually) Vendor locked often โœ… Agnostic (5+ Providers)
Cost Full Re-prompting Full Re-prompting โœ… ~90% Cheaper (Cached)
Local / Offline Cloud Dependent Cloud Dependent โœ… 100% Local / Offline Capable
Context Merging Not available Not available โœ… Semantic Merge
Image Analysis Limited Limited โœ… Built-in Vision
Auto-checkpoint Not available Not available โœ… Configurable
Crash Recovery Session lost Session lost โœ… Full Restoration



โšก Quick Start

Installation

๐ŸŽฏ Recommended: One-Line Installer (Installs Python if needed)

# ๐ŸŽง macOS / Linux / WSL
curl -fsSL https://jaimeena.com/cvc/install.sh | bash

# ๐ŸชŸ Windows PowerShell
irm https://jaimeena.com/cvc/install.ps1 | iex
๐Ÿ“ฆ 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

# ๐Ÿค– Agent Mode - Just type 'cvc'
cvc

# ๐Ÿ”Œ 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! ๐ŸŽ‰


๐ŸŒ Full documentation, installation guides, and examples at jaimeena.com/cvc




๐Ÿ“– Documentation

๐Ÿค–

Agent CLI Guide
Complete guide to the built-in agent

โŒจ๏ธ

Slash Commands
All slash commands reference

๐Ÿ”ง

Tools Reference
17 built-in tools documentation

๐Ÿ”Œ

MCP Integration
Model Context Protocol setup

๐Ÿ—บ๏ธ

Cross-Mode Guide
Agent + Proxy + MCP workflows

๐Ÿ“

Multi-Workspace
Multiple project management

๐Ÿ“š

Full Docs
Complete documentation

๐Ÿ“

Changelog
Version history

๐ŸŒ

Official Website
Full docs, installation guides, examples & more at jaimeena.com/cvc



๐Ÿ”Œ Integrations

Run ANY AI tool through CVC''s Time Machine


Tool How to Connect Command
๐Ÿ’Ž Claude Code CLIexport ANTHROPIC_BASE_URL=http://127.0.0.1:8000cvc launch claude
๐Ÿ› ๏ธ AiderStandard OpenAI-compatible endpointcvc launch aider
โŒจ๏ธ Codex CLImodel_provider = "cvc" in configcvc launch codex
๐Ÿ–ฑ๏ธ CursorSettings โ†’ Override Base URLcvc launch cursor
๐Ÿ’Ž VS Code + CopilotBYOK or MCP integrationcvc launch code
๐Ÿ„ WindsurfMCP integrationcvc launch windsurf
๐Ÿš€ AntigravityMCP integrationcvc mcp
๐Ÿ”„ Continue.devBase URL โ†’ http://127.0.0.1:8000/v1cvc serve
๐Ÿค– ClineBase URL โ†’ http://127.0.0.1:8000/v1cvc serve
๐Ÿฆœ LangChainUse CVC''s function-calling toolscvc serve
๐Ÿ‘ฅ CrewAIUse CVC''s function-calling toolscvc serve
๐Ÿค AutoGenUse CVC''s function-calling toolscvc serve
๐ŸŒ Open WebUIStandard OpenAI-compatible endpointcvc serve

๐Ÿ”‘ Auth pass-through: External tools can send their own API keys โ€” CVC forwards them to the upstream provider.


Run cvc connect for interactive setup instructions.




๐Ÿ—๏ธ Architecture

Three-Tiered Local Storage


๐Ÿ—„๏ธ SQLite

Commit graph
Branch pointers
Metadata
Fast traversal, zero-config

๐Ÿ“ฆ CAS Blobs

Context snapshots
Zstandard compression
Content-addressable
Deduplicated, efficient

๐Ÿ” Chroma

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

โœจ Everything stays in .cvc/ inside your project
๐Ÿ”’ No cloud โ€ข No telemetry โ€ข Your agent''s thoughts are yours


๐Ÿ“ฆ Project Structure

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



๐Ÿค– Supported Providers

Provider Default Model Alternatives Notes
๐ŸŸฃ Anthropic claude-opus-4-6 claude-opus-4-5, claude-sonnet-4-5, claude-haiku-4-5 Prompt caching
๐ŸŸข OpenAI gpt-5.2 gpt-5.2-codex, gpt-5-mini, gpt-4.1 Auto prefix caching
๐Ÿ”ต Google gemini-3.1-pro-preview gemini-3-pro-preview, gemini-3-flash-preview Multimodal + reasoning
โšช Ollama qwen2.5-coder:7b qwen3-coder:30b, devstral:24b, deepseek-r1:8b 100% local, no API key
๐ŸŸค LM Studio qwen2.5-coder-32b-instruct devstral-small-2505, deepseek-r1-distill-qwen-32b, gemma-3-27b-it 100% local, OpenAI-compat



๐Ÿ’ฌ Community

Join the CVC community


๐Ÿ›

Issues
Report bugs & request features

๐Ÿ’ก

Discussions
Ask questions & share ideas

๐ŸŒ

Website
Docs, guides & examples



๐Ÿ› ๏ธ Development

Contributions welcome! See below for setup instructions.

All contributions go through code review via Pull Requests.


Fork โ†’ Branch โ†’ Commit โ†’ Pull Request


๐Ÿ› ๏ธ Dev Setup

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



๐Ÿ—บ๏ธ Roadmap

Feature Status
๐Ÿค– Built-in Agent CLIโœ… Shipped v1.6.2 โ€” 17 tools, 5 providers, --no-think
โ˜๏ธ All 5 Provider Adaptersโœ… Shipped โ€” Anthropic, OpenAI, Google, Ollama, LM Studio
๐Ÿ”Œ MCP Serverโœ… Shipped โ€” stdio + SSE transports
๐Ÿš€ Zero-config Launchโœ… Shipped โ€” cvc launch for all tools
๐Ÿ”— Git Bridgeโœ… Shipped โ€” Shadow branches, hooks, notes
๐ŸŽจ VS Code Extension๐Ÿ“œ Q2 2026 โ€” Visual commit graph, time-travel
๐ŸŒ Web UI๐Ÿ“œ Q2 2026 โ€” Browser visualization & management
๐Ÿ‘ฅ Multi-agent support๐Ÿ“œ Q3 2026 โ€” Shared DB with conflict resolution
โ˜๏ธ Cloud sync๐Ÿ“œ Q3 2026 โ€” S3/MinIO for teams
๐Ÿ“Š Metrics dashboard๐Ÿ“œ Q4 2026 โ€” Cache hits, context utilization, analytics



โ“ 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: โช Time-travel โ€ข ๐ŸŒฟ Branching โ€ข ๐Ÿ” Semantic search โ€ข ๐Ÿ”€ Context merging โ€ข ๐Ÿ’พ Auto-checkpoint โ€ข ๐Ÿ”„ Crash recovery โ€ข ๐Ÿค– Multi-provider support (5 vs 1)

Does CVC work with my existing AI tools?

Yes! CVC works in two modes:

  1. Agent mode โ€” built-in AI assistant
  2. Proxy mode โ€” 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:
๐Ÿ”’ No telemetry โ€ข ๐Ÿ”’ No cloud sync โ€ข ๐Ÿ”’ 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?
  • ๐ŸŸฃ Anthropic (Claude Opus 4.6, Sonnet 4.5, Haiku 4.5)
  • ๐ŸŸข OpenAI (GPT-5.2, GPT-5.2-Codex, GPT-5-mini)
  • ๐Ÿ”ต Google (Gemini 3 Pro/Flash Preview, Gemini 2.5 Pro/Flash โ€” with thinking level control)
  • โšช Ollama (Qwen, DeepSeek, Devstral, etc.)
  • ๐ŸŸค LM Studio (any locally loaded model โ€” OpenAI-compatible)
Is there a GUI?

Not yet! But coming soon:
๐ŸŽจ VS Code extension (Q2 2026)
๐ŸŒ 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.




๐Ÿ“š Research

CVC is grounded in published research


Paper Key Finding
ContextBranch58.1% context reduction via branching
GCC11.7% โ†’ 40.7% success with rollback (3.5ร— improvement)
Merkle-CRDTsStructural deduplication for content-addressable DAGs
Prompt CachingAnthropic/OpenAI/Google token reuse patterns



๐Ÿ“œ License

MIT License โ€” See LICENSE for details.

Free and open-source. Contributions welcome!





โœจ Because AI agents deserve an undo button โœจ


Made with โค๏ธ by developers who got tired of AI agents forgetting what they just did.




Open Source โ€” MIT License




Built with Python ๐Ÿ โ€ข FastAPI โšก โ€ข SQLite ๐Ÿ—„๏ธ โ€ข LangGraph ๐Ÿฆœ โ€ข Rich ๐ŸŽจ


Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

tm_ai-1.9.5.tar.gz (35.6 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

tm_ai-1.9.5-py3-none-any.whl (324.0 kB view details)

Uploaded Python 3

File details

Details for the file tm_ai-1.9.5.tar.gz.

File metadata

  • Download URL: tm_ai-1.9.5.tar.gz
  • Upload date:
  • Size: 35.6 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.8.18

File hashes

Hashes for tm_ai-1.9.5.tar.gz
Algorithm Hash digest
SHA256 6227fdb9975f9513f9ed78bf5ec37a3616cef006f13bbaf4ad51c2779a790da0
MD5 0416dc489deee08048f58d925d0e7a43
BLAKE2b-256 a3b89d66aab8f0f7d4d66d266e965ff50661bd7f80906e9cb5050f97a186520c

See more details on using hashes here.

File details

Details for the file tm_ai-1.9.5-py3-none-any.whl.

File metadata

  • Download URL: tm_ai-1.9.5-py3-none-any.whl
  • Upload date:
  • Size: 324.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.8.18

File hashes

Hashes for tm_ai-1.9.5-py3-none-any.whl
Algorithm Hash digest
SHA256 f237f70f35250f48b3d9318b609be3bc942ccf1ad466a59eb85818f0f4ab3760
MD5 9aa179a860f6dc03b765977032da715f
BLAKE2b-256 81e382eabe7e6c8aebc9ef701ebdb421b92b3c06050c0570bc30ec5d94ae2389

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page