The whole project, exposed
Specs, stories, tickets, meeting notes, and git-linked status are queryable and writable by any MCP-compatible agent - not a limited subset.
Agents & MCP
Built on the protocol your coding agent already speaks - not a new thing to learn, a new thing it can finally use properly.
Specs, stories, tickets, meeting notes, and git-linked status are queryable and writable by any MCP-compatible agent - not a limited subset.
An agent assigns a task and gets a handle immediately. It checks status later instead of blocking on long-running work.
Before work is marked done, its output is checked against the Acceptance Criteria it was built from.
What an agent can read versus write is controlled explicitly per project and never left open by default.
One stop / full context