minddy

MCP serverEverything else

Minimal issue tracker: projects, objectives and trackable plans, written by agents over MCP.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use minddy

Install minddy

The server’s own address, for the clients that take one directly. Or connect ahel onceand every client you use reads it from one address, with the account kept on ahel rather than in each client’s config.

  • Claude Code

    claude mcp add --transport http minddy 'https://www.minddy.app/api/mcp'

    Run it once in your project, then open /mcp to approve any sign-in the server asks for.

  • Claude Desktop

    https://www.minddy.app/api/mcp

    Add a custom connector in Settings, paste this address, and approve the sign-in.

  • Cursor

    cursor://anysphere.cursor-deeplink/mcp/install?name=minddy&config=eyJ1cmwiOiJodHRwczovL3d3dy5taW5kZHkuYXBwL2FwaS9tY3AifQ==

    Open the link and Cursor adds the server at that address.

  • ChatGPT

    https://www.minddy.app/api/mcp

    In Settings, enable Developer mode, create an MCP app, and paste this address. Your plan and workspace must allow custom apps.

  • Codex

    codex mcp add minddy --url 'https://www.minddy.app/api/mcp'

    Run it once, then sign in with codex mcp login minddy if the server asks for an account.

From the project's README

As published by mangue-dev/minddy in README.md.

minddy is an open-source issue tracker for small product teams. It keeps the daily work in one place: projects, issues, objectives, saved views, collaborative pages, a public feedback board, and an MCP server that lets coding agents work with the same backlog as the team.

The application is built with Next.js, React, Tailwind CSS, and Supabase. It also includes optional macOS, Linux, and Microsoft Store Windows desktop shells and integrations for GitHub, GitLab, Stripe, and OpenRouter.

Product tour

Hand work to Numo

Start one Numo conversation from its page, the contextual floating button, or an issue action. Numo can answer with project context, update Minddy directly, or delegate repository work to a code worker in a server sandbox. Follow the work in the conversation, then review its linked pull request and diff without losing the context that started the request.

Keep durable knowledge beside the backlog

Write specifications, decisions, runbooks, and meeting notes in a nested project wiki. Pages support rich Markdown, issue mentions, attachments, history, and read-only sharing, so both people and connected agents can work from the same source of truth.

Connect customer feedback to delivery

Publish a feedback board where users can submit requests and vote. The team can triage posts, reply publicly, merge duplicates, and link feedback to issues so its public status follows the work through delivery.

Run locally

Requirements: Node.js 24 and pnpm 10 (the versions used by CI), plus a Supabase project for an interactive application.

corepack enable
corepack prepare pnpm@10.28.0 --activate
pnpm install --frozen-lockfile
cp .env.example .env
pnpm dev

.env.example documents every optional integration and the Supabase dashboard settings. At minimum, set MINDDY_PUBLIC_SUPABASE_URL, MINDDY_PUBLIC_SUPABASE_ANON_KEY, and SUPABASE_SERVICE_ROLE_KEY for a local application that can access its data. Never commit .env or production credentials.

The development command builds the agent-VM and page-markdown bundles before starting Next.js. See CONTRIBUTING.md before running code from an untrusted pull request: dependency installation, tests, and development commands execute repository code.

Minddy Cloud is the hosted and operated option from Minddy. Self-hosted is the same public core on infrastructure you control: no Minddy account, Stripe, PostHog, or managed provider is required. The edition guide explains the choice, data flows, responsibilities, and costs in one place. For a reproducible local or self-hosted Supabase bootstrap, see docs/self-hosting.md. On a fresh local clone, use pnpm bootstrap:supabase before pnpm dev instead of manually applying SQL in the Supabase dashboard. After installation, follow the self-hosted operations runbook for upgrades, coordinated Postgres and Storage backups, disaster recovery, and rollback. Release acceptance is recorded with the isolated clean-room self-hosting scenario. The official multi-architecture OCI image, its tag policy, and verification commands are documented in docs/container-image.md.

Common commands

pnpm dev          # run the web app locally
pnpm lint         # lint
pnpm typecheck    # type-check
pnpm test         # run the test suite
pnpm build        # production build
pnpm desktop:dev  # run the optional desktop shell

Architecture and deployment

Development follows a trunk-based branch contract: short-lived work branches merge by pull request into main, the integration branch and preview candidate. After CI and explicit production approval, automation fast-forwards production to the selected main SHA. There is no long-lived develop or release branch, and no human writes directly to production.

  • Web app: Next.js App Router, React, Tailwind CSS, and mangue-ui.
  • Data and auth: Supabase Postgres, Auth, Storage, and Realtime.
  • Numo and agent integration: Unified Numo conversations, server-sandbox code delegation, scheduled requests, and an OAuth 2.1 MCP endpoint for external agents.
  • Deployment: pnpm deploy is the interactive maintainer entry point. It detects whether to release the public core, deploy the Minddy Cloud web app, and publish desktop applications for macOS, Linux, and Windows, with automatic, all, custom, and Windows-only modes. The Windows-only mode reuses an existing core release and avoids starting macOS or Linux runners. The assistant waits for CI, requests an approved fast-forward from main to production, verifies the Vercel Production deployment, and only tags that deployed SHA. Builds and production secrets never come from the maintainer's machine. Self-hosters should adapt its production/Vercel conventions to their own hosting.

Public releases are distinct from deployments. Their SemVer/tag policy, artifacts, checksums, migrations, CI provenance, desktop distribution, and rollback procedure are documented in docs/releases.md. Linux installation, GPG verification, XDG paths, and update behavior are documented in docs/linux-desktop.md.

The CI workflow is the source of truth for checks. It runs the public-repository check, lint, typecheck, desktop bundle build, tests, and dependency audit. See CONTRIBUTING.md for contribution and review guidance, and SECURITY.md for the security model and vulnerability reporting.

Community and governance

  • Propose bugs and improvements through the repository's structured issue forms, and read CONTRIBUTING.md before opening a pull request.
  • Ask usage and self-hosting questions in GitHub Discussions. The boundaries between best-effort community help and commercial support are documented in SUPPORT.md.
  • Project roles, decisions, reviews, protected branches, and dependency policy are defined in GOVERNANCE.md.
  • Participation is subject to the Code of Conduct. Report vulnerabilities privately as described in SECURITY.md.

License

minddy is released under the GNU AGPL v3.0 only. If you operate a modified version for users over a network, you must offer those users the corresponding source code. See the licensing policy, including the limits on use of the minddy name and logos.

Advanced
Delivery
minddy MCP server → your ahel gateway (mcp.ahel.ai) → every connected AI client.
Catalog kind
mcp-server
Gateway key
io-github-mangue-dev-minddy
Source
github.com/mangue-dev/minddy
Hosted endpoint
https://www.minddy.app/api/mcp