Ashby MCP

Ashby's MCP server makes recruiting data available to AI clients so hiring context can be used outside the applicant tracking system. On the read side an agent can search and retrieve across candidates, applications, jobs, interviews, offers, notes, interview feedback, emails, referrals and hiring teams. That is the raw material for tasks recruiters actually repeat: summarizing where a pipeline stands, comparing a requisition against a headcount plan, drafting a weekly hiring update, or pulling the feedback history on a candidate before a debrief. Write access is deliberately narrow. An agent can add a note to a candidate, move a candidate between interview stages, and put a candidate forward for consideration on an open role. Offer approvals and other sensitive workflows stay in the product. Authentication uses user level OAuth rather than a shared service account, and every read and write respects the permissions that user already holds in Ashby, so a coordinator and a hiring manager see different results from the same prompt. Clients can be configured to require approval before an action is taken, which is worth enabling when the same session also has access to email. The server works with MCP compatible clients including ChatGPT, Claude and Cursor. It fits recruiting operations teams and hiring managers who want Ashby context inside the assistant they already use for writing and planning, rather than switching to the ATS for every lookup.