MCP Server

Connect WorkSignal to your AI coding tool and manage hiring in plain English. The WorkSignal MCP server translates natural language into API calls - no code required.

Setup

Choose your AI tool and follow the one-command setup:

Terminal

claude mcp add worksignal --transport stdio \
  -e WORKSIGNAL_API_KEY=ws_your_key_here \
  -- npx -y @worksignal/mcp-server

This adds WorkSignal to your global Claude Code MCP config. Restart Claude Code after setup.

To verify the connection, ask Claude:

"What WorkSignal tools do you have access to?"

"What's our current plan on WorkSignal?"

Option 1 - Terminal command

cursor mcp add worksignal "npx -y @worksignal/mcp-server" \
  -e WORKSIGNAL_API_KEY=ws_your_key_here

Option 2 - ~/.cursor/mcp.json

{
  "mcpServers": {
    "worksignal": {
      "command": "npx",
      "args": ["-y", "@worksignal/mcp-server"],
      "env": {
        "WORKSIGNAL_API_KEY": "ws_your_key_here"
      }
    }
  }
}

.vscode/settings.json (workspace) or User settings.json

{
  "mcp": {
    "servers": {
      "worksignal": {
        "command": "npx",
        "args": ["-y", "@worksignal/mcp-server"],
        "env": {
          "WORKSIGNAL_API_KEY": "ws_your_key_here"
        }
      }
    }
  }
}

Requires VS Code Insiders or a Copilot extension that supports MCP. Reload the window after saving.

Generic MCP config (Codex, Goose, opencode, etc.)

{
  "worksignal": {
    "command": "npx",
    "args": ["-y", "@worksignal/mcp-server"],
    "env": {
      "WORKSIGNAL_API_KEY": "ws_your_key_here"
    }
  }
}

The MCP server supports stdio transport. Add it to whatever config file your tool uses for MCP servers.

OAuth authentication

When you connect the MCP server using the commands above, it runs a standard OAuth 2.0 Authorization Code flow with PKCE automatically. Your tool will open a browser window to authorize access to your WorkSignal account.

Alternatively, you can authenticate using a direct API key (set WORKSIGNAL_API_KEY in the env block as shown above).

For full OAuth endpoint details, see Authentication.

Available tools

These tools are available to your AI coding tool after connecting the MCP server:

Tool	What it does
quick_post	Create and publish a job posting in one step
list_roles	List all roles - filter by status
get_role	Detailed role info with description and salary
create_role	Create a draft role
publish_role	Publish a draft role to your career page
pause_role	Pause a role to stop accepting applications
list_job_postings	See all postings on your career page
update_job_posting	Update a posting's title, description, or requirements
list_voice_screens	View screening results with scores
get_voice_screen	Full transcript, AI highlights, and red flags
list_candidates	List candidates, optionally filtered by role and stage
get_candidate	Candidate detail with applications and screen history
add_candidate	Add a candidate manually (referrals, inbound email, etc.)
advance_candidate	Move a candidate to the next pipeline stage
reject_candidate	Reject a candidate with a reason
hire_candidate	Mark a candidate as hired with optional start date
post_job_to_x	Post a job to X with chosen tone (professional, casual, bold)
get_channels	X and WhatsApp connection status and usage
list_x_posts	List X posts with engagement metrics
get_organization	Plan info, limits, and usage stats

Example prompts

Say these to your AI tool after connecting WorkSignal:

"Post a senior Rails engineer role on WorkSignal. Remote, $150-200K."

"Show me all completed voice screens for the frontend engineer role."

"What's our current plan and usage on WorkSignal?"

"Show me all candidates for the backend role who passed screening."

"Reject candidates #12 and #15 - not enough experience."

"Post the senior engineer job to X with a casual tone."

"Add my friend Sarah (sarah@example.com) as a referral for the frontend role."

"Pause the designer role - we have enough applicants."

API Reference Webhooks