Filippo-Venturini/ctxvault

Local memory infrastructure for AI agents. Store knowledge and skills in isolated vaults you compose, control and query.

60 ★10 forksPythonUpdated 2mo ago

What you need to know

Local memory infrastructure for AI agents with isolated, typed vaults: semantic memory for knowledge and skill vaults for procedural memory, exposed via CLI, HTTP API, and MCP.

Install

pip install ctxvault
uv tool install ctxvault for the MCP server

Usage

•Create a vault with ctxvault init and index documents
•Query semantically with ctxvault query
•Connect the MCP server (ctxvault-mcp) to Claude Desktop or Cursor
•Restrict vaults with --restricted and attach agents

Key features

✓Typed memory: semantic and skill vaults
✓Structural isolation with access control
✓Persistent memory across sessions
✓Human-observable via CLI and plain directories
✓Local-first, no cloud or telemetry
✓CLI, HTTP API, and MCP server interfaces

Best for

Developers who want a local, typed, and access-controlled memory layer for AI agents.

Caveats

⚠Requires Python 3.10+
⚠Restricted vaults require an --agent argument for API/MCP access

Platforms: macOS · Linux · WindowsClients: Claude Desktop · Cursor · Any MCP-compatible client

Reviewed 2026-08-11

Topics

agentic-aiai-agentschromadbclaudecli-tool embeddings fastapi knowledge-base langchainllmlocal-firstmcpmodel-context-protocolmulti-agentpersistent-memorypythonragsemantic-search vector-database

Stars: 60★
Forks: 10
Language: Python
License: MIT
Created: 2025-12-12
Last push: 2026-06-29