projectmem MCP

A local memory MCP server for coding agents: records issues, attempts, fixes and decisions and warns before repeating a failed approach

  • 880 ★ on the repository
  • Last commit
  • MIT
MCP serverOfficial

Extended access

Why this level

  • Runs a local server and stores project history on disk
All reasons and checks

riponcm/projectmem

Install

Manual install

pip install projectmem
pjm init

pjm init prints the MCP config block with your Python path; then restart the client.

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Runs a local server and stores project history on disk

This is third-party code. Review the repository files before installing.

What it does

projectmem gives a coding agent a memory and judgment layer for a project. It records issues, attempts, fixes and decisions, then warns the agent when it is about to repeat an approach that already failed. One MCP server serves every project at once, with no root or working-directory binding. Everything is stored locally, with no cloud and no telemetry. Besides the server there is a set of pjm commands for init, search, briefs, a dashboard and context capture. It works with Claude Code, Cursor, Antigravity and Codex, installs from PyPI, and pjm init prints a ready config block with your Python path.

Who it is for. For developers who want a coding agent to remember a project's history and not repeat past mistakes.

Good fit when

  • The agent repeats an approach that already failed
  • You want to store decisions and attempt history per project locally
  • You need shared memory across several projects

Not a fit when

  • The agent does not need project history
  • You cannot run a local server or store data on disk

Example request

Check the project memory whether we already tried this fix before repeating it

MCP

Transport
stdio
Authentication
not required

Limitations

The server and data are local, with no cross-machine sync out of the box. It needs Python, and the config points to its path. The memory is only as useful as how carefully attempts and decisions are recorded. MCP servers load only on a cold client start.

How to disable. Remove the projectmem server from your MCP config and uninstall the projectmem package.

FAQ

Does data go to the cloud?

No, everything is stored locally, with no cloud and no telemetry.

Do I set up the server per project?

No, one server with no root binding serves every project, and pjm init prints a ready config block.

Discussion

No questions or reviews yet.

Official

Local memory for agents: stores conversations and project files verbatim and retrieves them with semantic search over MCP

MCP serverExtended access60.5KRepository stars

An MCP server for a project memory bank: the agent stores and reads memory files through one service instead of inside the project

MCP serverExtended access925Repository stars
Official

Membase keeps one shared memory for AI agents that Claude Code, Cursor and Codex read through MCP, an SDK or the membase command

MCP serverExtended access141Repository stars

A local-first MCP memory server for agents: session history, codebase index and a concept graph in one SQLite database

MCP serverExtended access438Repository stars

An MCP server that gives an agent memory across sessions, projects and machines: decisions, dead ends, preferences and skills

MCP serverExtended access93Repository stars

Local project memory for agents: constraints, decisions and lessons live in an evidence-backed tree and return before a task.

MCP serverExtended access127Repository stars

Comparison with similar tools

Cards that share the main search query, and how they differ.

ToolOfficialAuthenticationRepository starsMain difference
projectmem (this card)yesnot required880
Personal Modelnonot required1,341local memory for Claude Code
The Agent Guildnonot required297guild keeps tasks
engrimno305local project memory for coding agents

README in short

The README describes projectmem as a local memory and judgment layer for coding agents: it records issues, attempts, fixes and decisions and warns before repeating a failed approach. One MCP server with no root binding serves every project, and data stays local with no cloud or telemetry. There are pjm commands for init, search and briefs. It installs from PyPI, and pjm init prints a config with your Python path. It works with Claude Code, Cursor, Antigravity and Codex. MIT licensed.

README badge

Report a problem