Skip to main content
MCP server CX Codex

pindoc — Claude Code MCP server: memory and setup

pindoc stores agent findings as typed, code-linked project records. Use it when a team needs reusable context across AI coding sessions.

Open on GitHub

What pindoc does

pindoc is a self-hosted memory system for projects built with AI coding agents. It turns durable findings into typed records such as decisions, analyses, debugging paths, tasks, verification notes, and glossaries. Records may link to commits, files, line ranges, URLs, and other Pindoc records.

Its MCP workflow exposes tools including pindoc.context_for_task, pindoc.artifact.propose, and pindoc.task.queue. A single /mcp endpoint can serve several projects, while each project-scoped call carries project_slug.

Docker Compose starts Postgres, pgvector, the Pindoc daemon, and the Reader SPA. The documented Docker route requires Docker 27+; the first run also needs outbound HTTPS so the EmbeddingGemma model and runtime can be cached. Configure the client with http://127.0.0.1:5830/mcp.

What it helps with

  • preserve architectural decisions for later agents
  • link debugging notes to commits and files
  • give a new agent project context before editing
  • inspect whether database migrations match the image
  • refresh stale semantic indexes

Example prompts for the agent

  • Save the authentication decision in Pindoc and link it to the current commit.
  • Ask Pindoc for task context related to src/auth.ts in project my-project.
  • Add this follow-up task to the Pindoc queue with its related files.

How to install pindoc

Install

The MCP registry has no launch command for it: see the repository README

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about pindoc

How do I install pindoc in Claude Code?

Clone it with git clone https://github.com/var-gg/pindoc.git, enter the directory with cd pindoc, then run docker compose up -d --build. Configure the client to use http://127.0.0.1:5830/mcp.

What does pindoc do?

It keeps reusable agent findings as typed project records connected to code. MCP tools guide context lookup, artifact proposals, and task handling, while the Reader is used for review.

What does pindoc need to run?

The Docker setup requires Docker 27+. The first launch also needs outbound HTTPS to cache EmbeddingGemma and its runtime.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor