Open Source · MIT Licensed

Your AI has
amnesia.
Let's fix that.

Persistent task and memory management for Claude Code, Cursor, Windsurf, and any MCP-compatible agent. Your AI remembers across every session.

Join the Waitlist Star on GitHub

Free forever for self-hosted. No credit card required.

Terminal — DoMCP

Session 1

> "Remember: auth uses JWT with RS256 and 24h token expiry"

Memory saved to project "backend-api"

Session 2 (new context window)

> "What auth approach are we using?"

Found memory: JWT with RS256, 24h token expiry

Applied to project "backend-api" — 2 related memories loaded

Your AI now remembers. Always.

Works with any MCP-compatible AI agent

Claude Code

Cursor

Windsurf

VS Code

Any MCP Client

See the difference

Without DoMCP

Forgets decisions made last session
Repeats the same mistakes
No continuity between conversations
Loses project context on restart
Starts from zero every single time

With DoMCP

Remembers every architectural decision
Learns from past context and patterns
Picks up exactly where you left off
Persistent memory across restarts
Builds knowledge over time

31+ tools. Zero forgotten context.

Todos, issues, memories, projects, cycles, and config — all through the Model Context Protocol.

Todos & Issues

Create, update, and track tasks and issues with priorities, severity, due dates, labels, and sprint cycles — just like Linear.

Persistent Memories

Store architectural decisions, debugging insights, and project context your AI recalls automatically in future sessions.

Project Context

Every session starts with full context — active todos, recent memories, project config, and active cycle loaded automatically.

Sprint Cycles

Plan and track work in time-boxed cycles. Assign todos to sprints and monitor progress across sessions.

Self-Hosted First

Run on your own infrastructure with Docker or npx. Your data stays on your machines. No account required.

Open Protocol

Built on the Model Context Protocol standard. Works with any MCP-compatible client, present and future.

60 seconds to an AI that never forgets

Three steps. No complex setup. No account required for self-hosted.

Install

One command. No account required.

npx @domcp/mcp-server@latest

Configure

Add to your AI agent's MCP config.

{
  "domcp": {
    "command": "npx",
    "args": ["@domcp/mcp-server"]
  }
}

Done

Your AI now has persistent memory.

✓ Memory saved to "backend-api"
✓ 3 related memories found

See it in action

Watch how DoMCP gives your AI persistent memory across sessions.

Session 1 — Monday 2:30 PM

Remember that we decided to handle Stripe webhooks with idempotency keys and store events in a processing queue before confirming.

Got it. I've saved this as a memory in your "payments-api" project. I'll reference this approach for all future Stripe webhook work.

Memory saved

New Session

Session 2 — Wednesday 10:15 AM

I need to implement the payment retry logic. How should we handle failed webhook deliveries?

Based on the approach you established — using idempotency keys with a processing queue — I'd recommend implementing exponential backoff retries that check the queue before reprocessing. This keeps your idempotency guarantees intact.

Recalled from memory

Simple, transparent pricing

Free forever for self-hosted. Pay only for cloud convenience.

Free

$0forever

Self-hosted. Your infrastructure, your data.

1 project
100 todos & 200 issues
50 memories
31 MCP tools
Docker & npx install
Community support

Popular

Pro

$10/month

Cloud hosted. Unlimited everything.

Unlimited projects
Unlimited todos & issues
Unlimited memories
Vector search (semantic)
Cloud hosted MCP server
Priority support

Team

$20/user/month

Collaboration features for teams.

Everything in Pro
Team sharing
Audit log
SSO (SAML)
Custom retention
Dedicated support

Stop re-explaining yourself
to your AI.

Give your AI persistent memory. Join the developers who are building with continuity, not repetition.