memcord

Memcord is a privacy-first, self-hosted MCP server that turns Claude conversations into a locally stored, searchable knowledge base. It organizes chat history by project slots, automatically summarizes conversations, and provides full-text search across past chats. The installation flow uses a Python-based runtime via uv, and the generator can output platform-specific MCP configurations so you can run Memcord alongside your preferred tooling. Once running, you can create or bind memory slots to your project directories, save conversations, and use built-in commands to read, summarize, and search across your memcord data. The server exposes a range of tools for project-aware memory management, including per-slot history organization, auto-summarization, and incremental merging of related conversations while preserving individual entries for context.

Prerequisites:

• Python 3.10+ installed on your system
• uv (Python package manager) installed (install via: curl -LsSf https://astral.sh/uv/install.sh | sh for macOS/Linux or the Windows PowerShell command in the README)
• Git

Install steps:

1. Clone the repository: git clone https://github.com/ukkit/memcord.git
2. Navigate into the repository: cd memcord
3. Create and activate a Python virtual environment using uv, then install in editable mode: uv venv && uv pip install -e .
4. Generate the MCP configuration for your platform: uv run python scripts/generate-config.py
5. Start Memcord using the recommended run command for Python/uv: uv run memcord
6. Optional: follow the installer’s documentation to configure IDEs and additional integrations.

Tips and common considerations:

• Memcord stores data locally for privacy; ensure your storage location has adequate space for your chat history.
• The MCP configuration generator creates per-platform config files; re-run generate-config.py if you change installation methods or environment.
• Auto-save hooks can be enabled to periodically save progress; see the config templates for details.
• If you upgrade Memcord, re-run the config generator to regenerate MCP configurations and adapt to any new options.
• Environment variables can be used to customize storage paths, log levels, and feature toggles; document any changes to ensure consistent behavior across restarts.