Connect your AI to Lever

What your AI can do with Lever

Once connected, these tools appear automatically in your AI assistant:

• list_postingsread-only

  List job postings. Filter by state, team, department, location, work type or tag, or group the results by one of those. Ask for `expand: ['owner', 'hiringManager']` to get the people behind each posting rather than their ids — Lever has no filter by person, so that is how you find someone's postings.

• get_postingread-only

  Get one job posting by id, with its content, team, location and owners.

• list_posting_usersread-only

  List the users who have access to a posting — its owner, hiring manager and followers.

• list_opportunitiesread-only

  List candidates (opportunities). Filter by posting, stage, tag, email, origin or date. `origin: 'referred'` is how you find candidates who came through employee referral. `expand: ['applications', 'stage', 'owner', 'contact']` saves a call per candidate.

• get_opportunityread-only

  Get one candidate (opportunity) by id.

• list_applicationsread-only

  List a candidate's applications — one per posting they applied to, with the answers they gave to that posting's questions.

• get_applicationread-only

  Get one application of a candidate in full.

• list_resumesread-only

  List a candidate's résumés. Each carries `parsedData`: past positions (company, title, summary, location, dates) and schools (institution, degree, field, dates), already extracted by Lever — so matching a candidate to a posting needs no file download.

• get_resumeread-only

  Get one résumé of a candidate, with its parsed positions and schools.

• list_filesread-only

  List the files attached to a candidate — cover letters, portfolios, take-home tests, anything a recruiter dropped on the profile. This returns the metadata only: name, size, type and upload date. Résumés have their own tool, list_resumes, and it is the better one to reach for because Lever parses those into structured positions and schools.

• get_fileread-only

  Get the metadata of one attached file: its name, extension, size and when it was uploaded. The contents are not returned — Lever serves those from a separate download route this connector does not expose.

• list_offersread-only

  List the offers made to a candidate: status, who created it, when, the salary and the other fields the offer form captured. The status is how you tell a candidate who was offered from one who signed — `signed` means accepted. Lever publishes no account-wide list of offers, so this is per candidate.

• list_feedbackread-only

  List the interview feedback forms filled in for a candidate, with every field and score. This is where an interview panel's actual judgement lives.

• get_feedbackread-only

  Get one feedback form filed on a candidate, with every field, the value the interviewer chose and when they completed it.

• list_feedback_templatesread-only

  List the feedback form templates of the account, with each field and its id. Required before create_feedback: the field ids of the values you submit come from here.

• get_feedback_templateread-only

  Get one feedback template with its fields, their type, their options and which stage it belongs to. Read it before update_feedback_template: that call replaces the template, so what you do not send back is what you lose.

• update_feedback_templatewrite · needs approval

  Rewrite a feedback template. This is account-wide and forward-looking: every interviewer who opens this form from now on sees what you leave here, so it is a change to the hiring process, not to one candidate. Lever replaces the template with what you send, so read get_feedback_template first and send back the fields you want to keep. Exactly one field must have type `score-system` — that is the overall rating Lever reports on.

• create_feedbackwrite · needs approval

  File a completed interview feedback form on a candidate. Lever records it under the user named in `performAs` and shows no sign that it was not typed by them — so only use this for someone who has told you what to record, never to invent an assessment. Take `baseTemplateId` and every field id from list_feedback_templates.

• update_feedbackwrite · needs approval

  Change the answers on a feedback form already filed on a candidate. Lever records the edit under the user named in `performAs` and shows no sign that it was not typed by them — so only use this for someone who has told you what to change, never to reword another interviewer's assessment. Only the fields you send are touched; take their ids from get_feedback. Setting `completedAt` is what marks a form as done.

• list_interviewsread-only

  List the interviews scheduled for a candidate, with their panel and interviewers.

• get_interviewread-only

  Get one scheduled interview: its panel, date, duration, location, the interviewers and the feedback form each of them owes. `feedbackForms` lists the forms already filed for it — an empty one is an interview nobody has written up yet.

• list_panelsread-only

  List the interview panels of a candidate. A panel is one scheduled block — an on-site day, a phone screen — holding the interviews it is made of, their date and their interviewers. Use it to see the shape of a candidate's process; use list_interviews for the individual slots.

• get_panelread-only

  Get one interview panel with every interview it contains, its stage, its timezone and who cancelled it if it was cancelled.

• list_referralsread-only

  List the referral forms filled in for a candidate. The referrer appears as a form field (often 'Name of referrer'), not as a Lever user — so counting who refers most means grouping on a typed name, and two spellings of one person will not merge. Find the candidates worth asking about with list_opportunities and `origin: 'referred'`; Lever publishes no account-wide list of referrals.

• get_referralread-only

  Get one referral form with every field the referrer filled in, and when they filled it.

• list_opportunity_notesread-only

  List the notes left on a candidate — the running commentary recruiters and managers add outside the formal feedback forms.

• get_opportunity_noteread-only

  Get one note thread on a candidate, with each of its comments, who wrote them and when.

• add_notewrite · needs approval

  Add a note to a candidate. Notes are visible to everyone with access to the candidate. Give `performAs` to record it under a specific user; without it Lever attributes the note to the account the API key belongs to. A note added here can be rewritten later with update_note, which is not true of notes typed into Lever itself.

• update_notewrite · needs approval

  Rewrite a note on a candidate. Two things make this narrower than it looks. Lever only accepts it on notes that were created through the API, so a note somebody typed into Lever cannot be edited here and the call is refused. And it replaces the whole thread: every comment you want to keep must be sent again, so read get_opportunity_note first and send back what should survive.

• list_stagesread-only

  List the pipeline stages of the account.

• get_stageread-only

  Get one pipeline stage by id — its name, so a stage id coming back from a candidate or an interview can be read as the step it actually is.

• list_usersread-only

  List the Lever users of the account — recruiters, hiring managers, interviewers. Their ids are what performAs and the posting owners refer to.

• list_archive_reasonsread-only

  List the reasons an application can be archived for — why candidates were turned down, and which of those reasons count as hired.

Example prompts

Just ask your AI in plain language — it picks the right Lever tool:

• “List job postings. Filter by state, team, department, location, work type or tag, or group the results by one of those. Ask for `expand: ['owner', 'hiringManager']` to get the people behind each posting rather than their ids — Lever has no filter by person, so that is how you find someone's postings — from Lever, in plain language.”
• “Get one job posting by id, with its content, team, location and owners — from Lever, in plain language.”
• “List the users who have access to a posting — its owner, hiring manager and followers — from Lever, in plain language.”
• “List candidates (opportunities). Filter by posting, stage, tag, email, origin or date. `origin: 'referred'` is how you find candidates who came through employee referral. `expand: ['applications', 'stage', 'owner', 'contact']` saves a call per candidate — from Lever, in plain language.”

How it works

1. Create your Proxoria account and add your MCP endpoint to Claude or ChatGPT.
2. Connect Lever with your API key — each user connects their own account, credentials encrypted at rest.
3. Your AI instantly gets the Lever tools. Write actions wait for your approval in the dashboard before anything changes.

Frequently asked questions

Does Lever have an official MCP server?

You don't need one. Proxoria exposes Lever to your AI through the Model Context Protocol, so you can use it from Claude, ChatGPT, Mistral or any MCP-compatible client — no official Lever MCP server required.

What can my AI do with Lever through Proxoria?

It gets 33 tools: list_postings, get_posting, list_posting_users, list_opportunities, get_opportunity, list_applications, get_application, list_resumes, get_resume, list_files, get_file, list_offers, list_feedback, get_feedback, list_feedback_templates, get_feedback_template, update_feedback_template, create_feedback, update_feedback, list_interviews, get_interview, list_panels, get_panel, list_referrals, get_referral, list_opportunity_notes, get_opportunity_note, add_note, update_note, list_stages, get_stage, list_users, list_archive_reasons. For example, list job postings. filter by state, team, department, location, work type or tag, or group the results by one of those. ask for `expand: ['owner', 'hiringmanager']` to get the people behind each posting rather than their ids — lever has no filter by person, so that is how you find someone's postings.

Is the Lever connector read-only?

No — it can also make changes. Write actions go through Proxoria's confirmation guardrail by default: your AI proposes them, and you approve each one from your dashboard before anything is written to Lever.

How do I connect Lever to my AI?

Create a Proxoria account, connect Lever with your API key, and add your Proxoria MCP endpoint to Claude or ChatGPT. Your AI instantly gains the Lever tools.