Workable MCP Integration

Weldable's Workable MCP integration connects your AI agents to your Workable hiring platform. Your agent can query open jobs, read candidate profiles, check application statuses, and pull interview data, all through natural language commands. No dashboard navigation, no manual exports.

Workable has spent nearly a decade building AI into its recruiting stack, trained on data from over 260 million candidates. Connecting it to an external AI agent through Weldable adds a new layer: your agent can combine Workable data with information from other tools, run cross-platform workflows, and take action on recruiting insights without switching between applications.

Use cases

Candidate pipeline monitoring

Your agent checks Workable for all active jobs and reports on pipeline health: how many candidates are at each stage, which stages have the longest average dwell time, and where candidates are dropping off. It sends a daily summary to the recruiting team's Slack channel. If any role has zero candidates in the screening stage, it flags the posting for review.

Interview scheduling coordination

When a candidate advances to the interview stage in Workable, your agent pulls their profile and availability notes, then creates calendar events for each interviewer on Google Calendar. It includes the candidate's resume link and the interview scorecard template in the event description. Recruiters get the scheduling done without the back-and-forth emails.

Hiring velocity reporting

Your agent queries Workable for all hires made in the past quarter, calculates time-to-fill for each role by department, and compares the numbers against your targets. It formats the data into a table and posts it to your leadership channel. Roles that exceeded the target by more than 10 days get a callout with the specific bottleneck stage identified.

Source effectiveness analysis

Your agent pulls candidate source data from Workable and calculates conversion rates from application to offer by source. It identifies which job boards, referral programs, or sourcing channels produce candidates that actually make it through the process, not just applications. The analysis runs weekly and updates a shared Google Sheet.

Stale job detection

Your agent scans Workable for job postings that have been open for more than 60 days with fewer than three candidates past the screening stage. It notifies the hiring manager and recruiting lead with a recommendation to refresh the job description, adjust compensation, or expand sourcing channels.

How it works

Connect your Workable account through Weldable's OAuth flow. Weldable requests the API scopes needed to read jobs, candidates, and interview data. Authentication tokens refresh automatically in the background.

Once connected, describe your request in plain English. Say "list all candidates who applied to the frontend engineer role in the last two weeks" and Weldable maps that to the correct Workable API calls, handles pagination, and returns the results. Your agent can then summarize, filter, or act on that data using any connected integration.

Tips

Workable's API uses subdomain-based authentication. Your Workable account has a unique subdomain (e.g., yourcompany.workable.com), and API calls are scoped to it. Make sure the correct subdomain is configured during setup, or requests will fail silently.

Candidate stages vary by pipeline. Workable lets you customize hiring pipelines per job. Your agent should query the pipeline configuration for each job before mapping stage names, since "Technical Interview" in one pipeline might be "Code Review" in another.

Workable's disqualification reasons are structured. When your agent processes rejected candidates, use the structured disqualification reason field rather than free-text notes. This makes aggregate reporting on rejection patterns much more reliable.

Rate limits are generous but still apply. Workable's API allows a reasonable request volume, but batch operations that touch hundreds of candidates should include delays between pages. Check the response headers for rate limit status to avoid hitting the ceiling during high-volume reporting runs.