xChuCx/agent-memory

Overview

A local, searchable project memory system for AI coding agents. It uses Markdown as the source of truth and provides an MCP interface for safe structured updates. No cloud services are required.

Best for

Best for
Developers who need persistent project memory for AI coding agents without cloud reliance.

Use cases

• Store project context and documentation as markdown files
• Enable AI coding agents to retrieve relevant project memory through search
• Update memory with structured updates from agent sessions

How to use

Install

npx -y @xchucx/agent-memory

Tools exposed

• agent-memory

Tested with

Claude Code, Cursor, AGENTS.md / Codex, Gemini CLI

Example client config

{\n  "mcpServers": [\n    {\n      "name": "agent-memory",\n      "command": ["agent-memory", "mcp", "--root", "${CLAUDE_PROJECT_DIR:-.}"]\n    }\n  ]\n}

Notes

A local, searchable project memory system for AI coding agents. It uses Markdown as the source of truth and provides an MCP interface for safe structured updates. No cloud services are required.

4 stars on GitHub. Last updated 2026-06-05. Licensed Apache-2.0.

Use cases

• Store project context and documentation as markdown files
• Enable AI coding agents to retrieve relevant project memory through search
• Update memory with structured updates from agent sessions

Pros

• Local storage ensures data privacy and no cloud dependency
• Markdown source of truth is easy to read, edit, and version control
• MCP interface enables seamless integration with AI coding agents

Cons

• Requires local setup and ongoing management
• No built-in versioning or collaboration features
• Markdown-based storage may not suit complex structured data

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.