Ashby MCP Server (Beta)

the ashby model context protocol (mcp) server lets ai models and agents access and interact with your ashby data through a standardized interface connect it to claude, chatgpt, cursor, or any other mcp compatible tool to query your recruiting pipeline, prepare for interviews, and take actions in ashby—all without leaving your ai client prerequisites an org admin must enable the mcp server toggle before users in your organization can connect navigate to admin > organization setup > opt in features and turn on mcp server permissions & access the mcp server is available on all ashby plans foundations legacy plus plus enterprise ✅ ✅ ✅ ✅ the mcp server is not available to analytics only organizations org admins can enable and manage the mcp server toggle in admin settings once enabled, elevated access users in your organization can connect their own accoun t only data visible based on the user's ashby permissions level will be available when connected to the mcp server authentication the mcp server authenticates each user individually via oauth every user completes their own oauth flow during setup, so the server only returns data that user can already see in ashby a user not involved in a hiring process cannot query candidates or roles they do not otherwise have access to set up your mcp client the ashby mcp server is client agnostic and works with any mcp compatible tool the server url is https //mcp ashbyhq com/mcp/v1 https //mcp ashbyhq com/mcp/v1 select your client below for setup instructions chatgpt currently, chatgpt can be connected to ashby as a custom app in developer mode this requires a chatgpt plan that supports custom connectors if your organization is on an enterprise account, custom apps may be restricted and you may need to ask your it team to add the ashby custom app for you in chatgpt, open settings > apps > advanced settings and turn on developer mode this allows you to add unverified connectors go back to apps and select create app in the new app dialog, fill in the following name ashby connection select server url and enter https //mcp ashbyhq com/mcp/v1 https //mcp ashbyhq com/mcp/v1 authentication select oauth select "i understand and want to continue" to acknowledge the custom server warning, then select create when prompted, select sign in with ashby and complete the oauth flow to connect your workspace claude currently, claude can be connected to ashby as a custom app in developer mode if your organization is on an enterprise account, custom connectors may be restricted—ask your it team to add the ashby connector for you open settings > connectors select add custom connector and enter the server url https //mcp ashbyhq com/mcp/v1 https //mcp ashbyhq com/mcp/v1 complete the oauth flow to connect your ashby workspace cursor and other ides open cursor settings > tools & mcps > new mcp server paste the following into your configuration { "mcpservers" { "ashby" { "url" "https //mcp ashbyhq com/mcp/v1" } } } save and restart cursor when you use an ashby tool for the first time, complete the oauth flow to connect your workspace available tools once you've connected the ashby mcp server, just describe what you want in plain language and your mcp client will pick the right tools and run them for you below is the list of available tools ashby adds more tools over time, including additional write actions to request a tool, use the submit mcp feedback tool described below mcp tool inputs and outputs may change without notice if you need a stable contract for either, build against our public api https //developers ashbyhq com/reference/introduction instead finding people and records search records by name — resolves a name to a record supports candidates, jobs, applications, and email templates this is the starting point for most questions, such as "find the candidate jordan lee" or "pull up the senior designer job " filter records — finds records matching specific criteria, and can return a total count instead of a list supports candidates, applications, jobs, job postings, openings, scheduled interviews, and offers for example "how many candidates are in the interview stage for the sales role" or "show me applications that involve me " describe object fields — a behind the scenes helper that discovers which fields to filter on for a given record type before calling filter records you do not invoke this directly, but it is what lets filtering work reliably across all supported record types get record details — pulls the full details of records you have already located supports jobs, job postings, applications, email templates, openings, scheduled interviews, and offers get candidate — pulls information from a candidate's profile resume, interview feedback, emails, texts, notes, interview transcripts, referrals, follow ups, form submissions, and more use this for questions like "summarize this candidate's background" or "did this candidate mention anything about relocation?" interviews and feedback get upcoming interviews — lists your upcoming interviews with candidate names, job titles, times, and links ask "what interviews do i have this week?" get interview details — pulls a detailed prep briefing for a specific interview candidate background, resume summary, the panel, and the hiring team use this right before an interview get interview plan — shows the stages a candidate moves through for a given job or application get submitted feedback — pulls up scorecard feedback you personally submitted for a candidate or interview this only shows your own feedback, not other interviewers' pipeline and your to do list get job pipeline — shows how many candidates are at each stage of a job's pipeline, and can compare multiple jobs at once ask "how is the pipeline looking for the engineering roles?" get pending tasks — lists your pending tasks in ashby scorecards you owe, approvals waiting on you, follow ups, candidate reviews, and survey requests ask "what do i need to do today?" making changes create candidate — creates a new candidate with name and contact details add note to candidate — drafts a note on a candidate's profile for you to confirm before saving can also reply to an existing note change application stage — moves a candidate's application to a different stage in the interview process for that job consider candidate for job — adds a candidate to a job by creating a new application at the first stage of that job's interview process getting help with ashby search product docs — answers questions about how to use ashby by searching the product documentation and returning an answer with links to relevant articles use this for "how do i" questions, such as "how do i set up an approval chain?" or "how do i create a job posting?" giving feedback submit mcp feedback — sends feedback to the ashby team if the mcp server hits a limitation or you have a suggestion if something is not working or you wish the server could do something it cannot, use this tool to let us know rate limits ashby mcp enforces two separate rate limits to keep the service reliable requests 120 per minute, per auth token tool calls 120 tool budget units per minute, per user and organization pair most tools cost 1 unit per call filter records costs 2 units what to do if you're rate limited most workflows won't hit these limits mcp clients typically make tool calls at a rate that's within these limits as they work through a task if you do see a rate limit error, prompt your mcp client to run fewer operations in parallel against ashby mcp, or wait a moment before retrying faqs can the mcp server access data i cannot see in ashby? no the mcp server uses user level oauth, which means it authenticates as you and inherits your existing ashby permissions it cannot return candidates, jobs, or other records you do not already have access to in ashby do other users see the data i query through the mcp server? no each user connects their own account through a separate oauth flow queries and responses are scoped to your individual session and your ashby permissions when will official connectors for claude and chatgpt be available? official connector submissions for both claude and chatgpt are currently in review when approved, connecting ashby will not require developer mode or manual setup steps check the ashby product updates page for the latest status can i use the mcp server with a client not listed here? yes the ashby mcp server is client agnostic and works with any mcp compatible tool point your client to https //mcp ashbyhq com/mcp/v1 https //mcp ashbyhq com/mcp/v1 and complete the oauth flow when prompted what should i do if a tool is not working as expected? use the submit mcp feedback tool in your ai client to send feedback directly to the ashby team include as much detail as possible about what you asked and what the server returned