Repolode
Submit
MemPalace

MemPalace/mempalace

Local-first AI memory system that stores conversation history verbatim and retrieves it with semantic search, exposed as an MCP server and CLI.

Open-source alternative to:Mem0

View on GitHubWebsite
Stars
59,652
Forks
7,569
Open issues
826
Last commit
6 hours ago

What is it?

MemPalace stores conversation history and project files as verbatim text and retrieves them with semantic search. It does not summarize or paraphrase content. The index is organized into wings (people and projects), rooms (topics), and drawers (original content), so searches can be scoped to a specific area instead of a flat corpus.

The default storage backend is ChromaDB, and the backend interface is pluggable. Milvus, Qdrant, pgvector, and SQLite-based backends are also offered as opt-in options. Embeddings can be computed locally or through any OpenAI-compatible /v1/embeddings endpoint. Nothing leaves the machine unless you configure it to.

MemPalace ships a CLI for mining, searching, and auditing a palace, plus an MCP server with 47 tools for reads, writes, knowledge-graph operations, and agent coordination. Auto-save hooks are available for Claude Code, Codex CLI, and Cursor IDE.

Key features

  • 01Stores conversation and project content verbatim and retrieves it by semantic search.
  • 02Organizes content into wings, rooms, and drawers so searches can be scoped.
  • 03Pluggable storage backends including ChromaDB (default), Milvus, Qdrant, and pgvector.
  • 04Temporal knowledge graph with validity windows, backed by local SQLite.
  • 05MCP server with 47 tools, runnable via pip, uv, or Docker.
  • 06Auto-save hooks for Claude Code, Codex CLI, and Cursor IDE.

Use cases

  • Giving a coding agent long-term memory of past sessions and project decisions.
  • Searching old chat transcripts to recover why a design choice was made.
  • Running a private, local memory store without sending data to a cloud API.
  • Sharing a memory hub across a team through the provided server deployment.

Quick start

  1. 1

    Install the CLI with uv

    uv tool install mempalace
  2. 2

    Initialize a project palace

    mempalace init ~/projects/myapp
  3. 3

    Mine project files into the palace

    mempalace mine ~/projects/myapp
  4. 4

    Search the palace

    mempalace search "why did we switch to GraphQL"

Things to know

  • Native Termux installation is not supported; Android users need a Debian PRoot container.
  • The GPU Docker image is x86_64-only and is not published, so it must be built locally.
  • The first embedding call downloads a model (about 80 to 300 MB) and needs network access.

Topics

#ai-memory#semantic-search#vector-database#chromadb#mcp#llm#rag#python#ai#memory

FAQ

What is mempalace?
Local-first AI memory system that stores conversation history verbatim and retrieves it with semantic search, exposed as an MCP server and CLI. MemPalace stores conversation history and project files as verbatim text and retrieves them with semantic search. It does not summarize or paraphrase content. The index is organized into wings (people and projects), rooms (topics), and drawers (original content), so searches can be scoped to a specific area instead of a flat corpus.
Is mempalace free to use? What license does it have?
mempalace is open source under the MIT license. Check the license terms before using it in commercial products.
Is mempalace still maintained?
Actively maintained. The last commit to mempalace was on Oct 20, 2026.
What can I use mempalace for?
Giving a coding agent long-term memory of past sessions and project decisions. Searching old chat transcripts to recover why a design choice was made. Running a private, local memory store without sending data to a cloud API. Sharing a memory hub across a team through the provided server deployment.
How do I install mempalace?
Install the CLI with uv: uv tool install mempalace Initialize a project palace: mempalace init ~/projects/myapp Mine project files into the palace: mempalace mine ~/projects/myapp Search the palace: mempalace search "why did we switch to GraphQL"
Where can I get mempalace?
The source code and installation instructions are on GitHub: https://github.com/MemPalace/mempalace

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects