EXPONENTIAL
An issue tracker a team runs its daily work on, built around AI agents connected over MCP. An agent analyses an issue and proposes options, a person makes the call, and commits, deploys and change notes land back on the issue. Open source, with a live demo.
- A kanban board, sprints with a review, sub-issues and blocking relations, with changes live in every open tab (Postgres LISTEN/NOTIFY and Server-Sent Events).
- An MCP server lets agents read and update issues on per-user tokens, and every agent action is marked in the history.
- Analysis comes in sections with decisions and a recommended option that a person picks; the choice goes into the prompt the agent starts work from.
- Commits link to issues by ID, deploy events arrive from a webhook, GitHub Deployments or Coolify, and human and technical change notes add up to the sprint review.