DoneThat
Connect an agent to privacy-first work logs, summaries, and long-term memory so shipping stays coherent across sessions.
Overview
DoneThat is an Operate-phase MCP server that gives AI clients privacy-first work tracking, summaries, coaching, and long-term memory via https://mcp.donethat.ai/.
What is this MCP server?
- Hosted streamable HTTP MCP at mcp.donethat.ai—no local server to maintain
- Privacy-first work tracking with summaries and reports for solo builders
- Coaching-oriented insights on how work is progressing
- AI-ready long-term memory so agents retain context across days and projects
- Remote-only integration suited to Claude Code, Codex, and Cursor MCP clients
- MCP registry version 1.0.0
- 1 production remote endpoint: https://mcp.donethat.ai/
- Transport: streamable-http
What problem does it solve?
Agent chats forget what you shipped last week, so you repeat planning and lose thread on solo projects.
Who is it for?
Indie builders who want agent-aware work logs and memory without feeding everything into a public SaaS tracker.
Skip if: Teams that need full Jira-style issue hierarchies, time billing, or on-prem-only data with no remote MCP.
What do I get? / Deliverables
Your agent can read and update structured work history and memory so status, priorities, and coaching stay consistent across sessions.
- Connected remote MCP endpoint for work tracking and memory tools
- Agent-accessible summaries and reports from your tracked work
- Persistent coaching-oriented context across agent sessions
Recommended MCP Servers
Journey fit
Canonical on Operate because the core value is sustaining momentum and reflection after you are already building—not one-off codegen. Iterate is where solo builders reconcile what shipped, what stalled, and what to do next; DoneThat is built for that loop with coaching and reports.
How it compares
Remote productivity-memory MCP, not a repository skill or a generic GitHub integration.
Common Questions / FAQ
Who is DoneThat for?
Solo and indie builders using Claude Code, Cursor, or Codex who want privacy-oriented work tracking and AI-ready memory connected through MCP.
When should I use DoneThat?
Use it when you are iterating in production or between sprints and need summaries, reports, and durable context the agent can query without re-pasting notes.
How do I add DoneThat to my agent?
Add the remote MCP server URL https://mcp.donethat.ai/ with streamable HTTP in your client’s MCP settings, then complete any auth flow the host requires.