Greenhouse MCP Integration

Weldable's Greenhouse MCP integration gives your AI agents direct access to your Greenhouse recruiting data. Your agent can read job postings, query candidate pipelines, pull interview scorecards, and manage applications through natural language, all without writing custom API code or managing authentication tokens.

Greenhouse is one of the most widely adopted applicant tracking systems, used by thousands of companies to run structured hiring processes. Connecting it to an AI agent through Weldable means your recruiting operations data becomes actionable: your agent can answer questions, flag bottlenecks, and coordinate across tools in real time.

Use cases

Hiring pipeline health checks

Your agent queries Greenhouse for all open jobs and calculates the number of candidates at each stage. It flags roles where the top of funnel is thin (fewer than five candidates in "Application Review") or where candidates are stacking up in a single stage for more than a week. The results go into a Slack message to hiring managers each Monday morning, giving them a snapshot without opening Greenhouse.

Scorecard compliance tracking

Greenhouse supports structured interviewing through scorecards, but interviewers don't always fill them in promptly. Your agent checks for completed interviews with missing scorecards, then sends a direct reminder to each interviewer. It tracks compliance rates over time and includes them in weekly recruiting ops reports.

Cross-tool candidate coordination

When a candidate advances to the onsite stage in Greenhouse, your agent pulls their resume and interview history, creates a prep document in Google Docs with key talking points for each interviewer, and schedules the debrief meeting on Google Calendar. The recruiting coordinator sees the prep work already done when they check in the next morning.

Rejection reason analysis

Your agent pulls rejection data from Greenhouse over the past quarter, groups candidates by rejection reason and stage, and identifies patterns. If 40% of rejections at the technical screen cite "insufficient system design experience," your agent surfaces that to the sourcing team so they can adjust their outreach criteria.

Job posting performance

Your agent queries Greenhouse for active job postings and cross-references them with application volume over the past 30 days. Postings with low application rates get flagged, and your agent can suggest edits based on what's working for similar roles. The report lands in your recruiting channel without anyone pulling CSV exports.

How it works

Connect your Greenhouse account by providing an API key through Weldable's setup flow. Greenhouse uses its Harvest API with HTTP Basic authentication and granular permission scopes. Weldable stores your credentials securely and handles authentication on every request.

Tell your agent what you need in natural language. A request like "show me all candidates in the onsite stage for the product manager role" gets mapped to the right Greenhouse API endpoints. Weldable resolves job IDs, handles pagination for large result sets, and returns structured data your agent can summarize or act on.

Tips

Greenhouse permissions are granular, so scope them carefully. The Harvest API supports permission levels per object type (candidates, jobs, scorecards). Create a dedicated API credential for Weldable with only the access your agent needs. Read-only is fine for reporting workflows; read-write is needed if your agent will move candidates or update profiles.

Scorecards are tied to interview kits, not just candidates. When querying scorecard data, your agent needs both the application ID and the interview ID. Greenhouse's API returns scorecards nested under interviews, so a direct candidate lookup won't include them.

Use custom fields for agent metadata. Greenhouse supports custom fields on candidates and jobs. Use these to store flags your agent sets (like "pipeline-stale" or "source-quality-low") without cluttering the standard fields that recruiters rely on.

Pagination is cursor-based for most endpoints. Greenhouse returns a Link header with pagination URLs. Your agent should follow these cursors rather than guessing page numbers, especially when pulling large candidate lists or historical data.