Weldable's Lever MCP integration connects your AI agents directly to your Lever ATS account, giving them the ability to read candidates, manage pipeline stages, and act on recruiting data through natural language. Lever combines applicant tracking with candidate relationship management, and an MCP connection turns all of that data into something your agent can query and update without anyone clicking through the Lever UI.
Recruiting teams spend hours each week on pipeline hygiene: moving candidates between stages, leaving feedback notes, and pulling reports for hiring managers. With Weldable handling the API authentication and intent mapping, your agent takes on that operational work while recruiters focus on the conversations that actually close hires.
Use cases
Pipeline stage automation
Your agent monitors candidates sitting in a stage for longer than a configured threshold. When a candidate has been in "Phone Screen" for five days with no feedback submitted, the agent posts a reminder to the hiring manager in Slack and updates the candidate's tags in Lever to flag the delay. This keeps pipelines moving without recruiters manually auditing every opening.
Interview feedback collection
After each interview block, your agent checks Lever for completed interviews that are missing scorecard submissions. It messages the interviewers directly with a link to submit their feedback and a deadline. If feedback still hasn't arrived by end of day, the agent escalates to the hiring manager. The entire follow-up loop runs without a recruiting coordinator touching it.
Candidate source reporting
Your agent pulls candidate data from Lever, groups it by source (referral, inbound, sourced, agency), and calculates conversion rates through each pipeline stage. It formats the results into a weekly digest and posts it to your team's reporting channel. Hiring leaders get a clear picture of which sources produce interviews and offers, not just applications.
Offer stage coordination
When a candidate moves to the offer stage, your agent pulls their profile details from Lever, checks compensation benchmarks from a connected Google Sheet, and drafts a summary for the compensation committee. It then creates a calendar event for the approval meeting and notifies the recruiter once the meeting is scheduled.
Duplicate candidate detection
Your agent periodically scans recent applications in Lever and cross-references them against existing candidate profiles by email and name. When it finds a likely duplicate, it tags both profiles and notifies the recruiter with links to review. This prevents split candidate histories and ensures interviewers see the full picture.
How it works
Connect your Lever account through OAuth. Weldable requests the scopes needed to read and write candidate, opportunity, and feedback data. Tokens refresh automatically, so your agent maintains access without manual re-authentication.
Once connected, describe what you need in plain English. Say "find all candidates in the technical phone screen stage for the backend engineer role" and Weldable maps that to the correct Lever API calls, resolves posting IDs, and returns structured results your agent can act on.
Tips
Use Lever's tags for agent-driven workflows. Tags like "agent-flagged" or "stale-pipeline" give your agent a way to mark candidates for follow-up without changing their pipeline stage. This keeps the recruiting workflow clean while giving the agent a lightweight signaling mechanism.
Lever's archive reasons matter for reporting. When your agent archives a candidate, always include a specific archive reason. Lever's built-in analytics use these reasons to calculate pass-through rates. Generic reasons like "other" make your funnel data unreliable.
Opportunities and candidates are separate objects in Lever. A single candidate can have multiple opportunities (one per job they applied to). Make sure your agent queries at the opportunity level when working with pipeline data, or you'll get incomplete results for candidates who applied to more than one role.
Rate limits on the Lever API are per-user. Lever enforces rate limits at the API key level, typically 10 requests per second. If your agent runs batch operations like scanning all open candidates, add pagination and respect the rate headers to avoid throttling.
Works well with
Connect your agent to Lever
Connect your Lever account and start automating with AI agents in minutes. Free to use, no credit card required.