Orchestro is an open-source MCP server + web dashboard for Claude Code.
Think Trello for Claude Code — but with a built-in Scrum Master that keeps the board honest and agents that move the cards from goal → tasks → code as work happens.
• Website: orchestro.org
• Repo: github.com/khaoss85/mcp-orchestro
• License: MIT • Try time: ~2 minutes
Why I built it
Even with the best agent UX I’ve tried (Claude Code), I kept hitting the same walls:
- intent/decisions buried in prompts
- dependencies invisible until too late
- handoff goal → tasks → code drops context
- PMs and devs don’t see the same board
I wanted a thin, no-drama layer that keeps the plan visible and the execution honest—without killing speed.
What it does
- Board that actually executes: plan in Kanban, the MCP server drives the flow
- Scrum Master (auto): decomposes stories, sets dependencies, keeps lane hygiene
- Agents move cards: prep context-rich prompts for Claude and move work across columns
- Live status: you and your PM see the same board in real time
- Local-first: your data lives in your Supabase; no hardcoded secrets
Quick start
npx [@orchestro](/orchestro)/init
npm run dashboard # opens http://localhost:3000
(restart Claude Code, then ask: "Show me orchestro tools")
My ask (open source)
I shipped my first MCP here and want real users (especially heavy Claude Code folks) to break it and tell me what hurts.
- ⭐ If useful, star the repo
- 🐞 If not, open an issue (what blocked you? where did it feel slow?)
- 🔧 PRs and brutal feedback welcome
- 🗣️ If you run a team and try it, a 10–15 min call to debrief would help me prioritize
Who should try this
- Builders who live in Claude Code and want fewer invisible steps
- Solo founders & small teams doing vibe coding but needing a clean map for stakeholders
- PMs who want to see progress without spelunking through prompts
Roadmap (short)
- PM-friendly exports & notifications
- Velocity/estimation & lightweight metrics
- Slack/Teams hooks
If it helps you ship faster, awesome.
If it doesn’t, please tell me why so I can fix the right thing.
“Thin, no-drama layer” is exactly the right framing.
The adjacent problem I keep feeling is the approval boundary once local execution starts. Keeping the board honest is one part. The other is making sure repo work still passes through one visible plan and one real gate before anything actually runs.
Curious whether you think that belongs inside Orchestro or whether it wants to stay a separate layer.