Using the Workable MCP server to connect to AI assistant

The Workable MCP (Model Context Protocol) server lets you connect your AI assistant directly to your Workable account. Once connected, you can manage recruiting and HR tasks through natural language, asking your AI assistant to pull candidate lists, move applicants through the pipeline, approve requisitions, check time-off balances, and more, without leaving your AI tool.

This article covers what you can do with the Workable MCP server and how to get connected. To learn more, check our documentation article.

Before you begin

To use the Workable MCP server, make sure you have:

• An active Workable account.
• An AI client that supports MCP, such as Claude, Cursor, or any other MCP-compatible tool.

The Workable MCP server uses OAuth2 for authentication. You'll sign in with your existing Workable credentials when connecting - no separate API key is required for the initial setup.

Connecting the Workable MCP server

To connect the Workable MCP server to your AI client, follow these steps:

1. Open your AI client's MCP configuration settings. Depending on your tool, this may be a settings menu, a configuration file, or a connector marketplace.
2. Add a new MCP server using the following URL: https://mcp.workable.com/mcp
3. When prompted, sign in with your Workable account credentials to authorize the connection via OAuth2.
4. Once authenticated, your AI assistant will have access to your Workable data and can start performing actions on your behalf.

Important: The actions your AI assistant can take through MCP are governed by your Workable account permissions. It can access data and perform actions only if your user account is authorized to do so.

What you can do with the Workable MCP server

Once connected, your AI assistant has access to over 40 tools across your Workable account. Here's what you can do, organized by area:

Example prompts

Here are a few things you can ask your AI assistant once connected:

• "Show me all candidates in the final interview stage for the Senior Engineer role."
• "Move Jane Smith to the Offer stage in the Product Manager job."
• "What time-off requests are pending for this month?"
• "Create a new requisition for a Customer Support Specialist in Athens."
• "Approve the offer for Alex Johnson."
• "List all published jobs and their pipeline stage counts."

Common use cases

Candidate pipeline inactivity alerts

Candidates often get stuck in a hiring stage for weeks with no activity — no one notices, no one acts. This automation runs a daily or weekly scan across all open jobs, identifies any candidate who hasn't moved stages or had an event logged in a configurable number of days, and pushes a Slack alert to the relevant hiring manager.

How it works

The skill connects to Workable's API to pull all active candidates across all jobs. It then compares the date of the last recorded activity (stage change, comment, evaluation, email) against today's date. Any candidate exceeding your inactivity threshold — typically 7 or 14 days — is flagged.

An alert is then sent to the relevant hiring manager via Slack, summarising all stale candidates grouped by job. For example:

• "3 candidates in Phone Screen for Project Manager have had no activity in 14 days."
• "2 candidates in Technical Test for Senior Engineer haven't moved in 10 days."

Note: This is something Workable's native notifications don't do across all jobs at once in a summarised way. The value is the aggregated, cross-job visibility delivered directly to the person who needs to act.

Configuration

You can configure the inactivity threshold (e.g. 7, 10, or 14 days), which stages to monitor, and whether the alert goes to individual hiring managers or a centralised Slack channel. The scan can be scheduled to run daily or weekly depending on your hiring volume.

Open requisitions weekly status brief

When you have 10+ open requisitions — several with planned start dates already in the past — it's easy to lose track of which ones are most urgent and whether hiring has actually started for each. This automation generates a weekly digest that connects the headcount approval side (requisitions) with actual hiring activity.

What the digest includes

Each week, the skill pulls all open requisitions from Workable and ranks them by how overdue they are, from most to least. For each requisition, the digest shows:

• The requisition name and the owner responsible for it.
• The original planned start date and how many days overdue it currently is.
• Whether a linked job exists in Workable, and if so, how many active candidates are currently in the pipeline.
• Requisitions with no linked job — meaning headcount has been approved but hiring hasn't started — are highlighted prominently.

Note: The digest is designed as a quiet nudge, not a loud alarm. It surfaces the information to the right people at a regular cadence so that overdue requisitions don't go unnoticed week after week.

Delivery

The brief can be sent to an individual (e.g. the Head of Talent or Finance lead) or posted to a dedicated Slack channel. It is typically scheduled for Monday mornings so that owners can plan follow-up actions for the week ahead.

Weekly hiring activity digest

Instead of a dry numbers summary, every Friday this skill pulls the week's pipeline movement and writes it as a short, readable narrative — sent to a Slack channel — that people can actually read and act on.

What makes it different

Standard ATS reports show tables and counts. This digest tells the story of the week in hiring. Rather than "7 candidates advanced", it might say:

• "A strong week for Engineering — 3 candidates advanced to interview, and one offer is now pending for the Senior Backend Engineer role."
• "The Project Manager role in Boston hasn't seen pipeline movement since February and may need attention."
• "Two new applications came in for Relationship Banker — both have been sitting in Applied for 6 days with no action."

How it works

Each Friday, the skill queries Workable's API for all pipeline events from the past 7 days — stage changes, offers sent, interviews scheduled, rejections. It then uses Claude to synthesise those events into a narrative digest, flagging both positive momentum and potential concerns.

Note: The tone is intentionally journalistic, not bureaucratic. The goal is a digest that a hiring manager or leadership team will actually read on a Friday afternoon — not one they'll skip because it looks like a spreadsheet.

Delivery

The digest is posted to a Slack channel of your choice every Friday at a scheduled time. You can configure which jobs or departments to include, and whether to mention specific individuals (e.g. the hiring manager for a stalled role).

Pending candidate communication

Every candidate who reaches the Applied stage and hasn't received any communication or stage movement in 5 or more days is a candidate having a poor experience — and potentially sharing that experience publicly. This automation makes candidate communication dead zones visible with zero manual effort.

How it works

On a weekly schedule, the skill scans all candidates currently in the Applied stage and checks when the last outbound communication or stage event was logged against their profile. Any candidate who has been waiting for 5+ days without contact is flagged as a "communication dead zone."

A summary is sent to the relevant recruiter listing all flagged candidates, along with a pre-drafted acknowledgement email for each one — ready to review and send in a single action. For example:

• The recruiter receives a Slack message: "4 candidates in Applied for Customer Success Manager have had no communication in 5+ days."
• A draft email is prepared for each candidate acknowledging their application and giving them a realistic timeline.

Note: Candidate experience is one of the most undertracked metrics in ATS tools. A single unacknowledged application can result in a negative Glassdoor review or a declined offer later in the process. This automation surfaces the risk early — before it becomes a reputation problem.

Configuration

You can configure the inactivity threshold (default: 5 days), which stages to monitor beyond Applied, and whether draft emails are included in the alert or sent to a separate inbox for review. The scan can be run daily for high-volume roles or weekly for standard pipelines.

Contractor end date & renewal notifications

For any employee with a contract or employment expiration date populated in Workable, this scheduled automation sends a timely notification to their manager — giving enough lead time to plan for offboarding, extension, or conversion before the end date arrives.

How it works

The skill runs on a scheduled basis (daily or weekly) and scans all employee records in Workable for populated contract end date fields. For each contract expiring within your configured notice window — for example, 30, 60, or 90 days — a notification is sent to the employee's direct manager via email or Slack.

The notification includes:

• The contractor's name, role, and department.
• The exact contract end date and the number of days remaining.
• A prompt to initiate one of three actions: extend the contract, begin offboarding, or convert to a permanent role.
• A link directly to the employee's profile in Workable for quick access.

Note: A second reminder can be configured to fire closer to the end date (e.g. 14 days out) for any contracts where no action has been taken. This prevents contractors from reaching their end date without any offboarding or extension steps being initiated — a common and avoidable operational gap.

Configuration

You can set multiple notice windows (e.g. 60 days, 30 days, 14 days), choose between email and Slack delivery, and specify whether notifications go to the direct manager only or also to HR or a People Ops channel. The skill can be scoped to specific employment types (e.g. contractors only, or all fixed-term employees).

FAQs

Can I connect multiple Workable accounts to the MCP server?

Yes. If you manage multiple Workable accounts, the MCP server supports connecting to more than one account.

Which AI assistants are compatible with the Workable MCP server?

Any AI client or tool that supports the Model Context Protocol (MCP) can connect to the Workable MCP server. This includes Claude, Cursor, and other MCP-compatible tools. Check your AI tool's documentation to confirm MCP support and to find configuration instructions specific to that client.

What level of Workable access do I need?

You need an active Workable user account. Once connected, the tools available to your AI assistant will reflect your individual Workable access level; it cannot access data or perform actions beyond what your account is authorized to do.

Is my Workable data secure when using MCP?

The Workable MCP server uses OAuth2 authentication, meaning your credentials are never stored by your AI client. Your AI assistant only has access to data within your Workable account, and all actions are performed under your user identity and permissions.

Can my AI assistant take actions in Workable, or only read data?

Both. The Workable MCP server supports read operations (listing jobs, viewing candidate profiles, checking time-off balances) as well as write operations (moving candidates, creating requisitions, approving offers, adding comments). Always review what your AI assistant is about to do before confirming actions that modify data.

I'm having trouble connecting. What should I do?

Make sure you're using the correct server URL (https://mcp.workable.com/mcp) and that you have access in Workable. If the issue persists, consult your AI client's MCP setup documentation or contact Workable Support.