Skip to main content
The Sentrion Model Context Protocol (MCP) server lets AI assistants search current and historical job postings using your Sentrion account. It exposes eight tools backed by the Sentrion API.

Connect to Sentrion

The server URL is https://mcp.sentrion.ai/mcp (Streamable HTTP). Connect by signing in with your Sentrion account, or with an API key if your client doesn’t support OAuth.

Sign in with your Sentrion account

Use this for clients with a Connect or Sign in button for remote MCP servers, such as Claude and ChatGPT.
1

Add the MCP server

Add a remote server (sometimes called a custom connector) named Sentrion with the URL https://mcp.sentrion.ai/mcp. Leave authentication settings empty.
2

Sign in

Your client opens a browser window. Sign in with the account you use for the Sentrion dashboard and approve access.
3

Verify the connection

Ask your assistant: “Check my Sentrion credit balance.” It should call get_credit_balance, which consumes no credits.
Tool calls use your account’s API key, so credits and billing work exactly as they do with API-key access. If you’ve never signed in to the dashboard, do that once first: your API key is created on your first dashboard sign-in.

Use an API key

Use this for clients that support a bearer token or custom Authorization header but not OAuth, and for scripts.
1

Get your API key

In the Sentrion dashboard, open Settings > API Keys and create a key.
2

Add the MCP server

Add a remote server in your client’s MCP settings using these values:Replace YOUR_API_KEY with your key. If your client has a dedicated bearer-token field, enter the key alone; the client adds the Bearer prefix.
3

Verify the connection

Ask your assistant: “Check my Sentrion credit balance.” It should call get_credit_balance, which consumes no credits.
Keep your API key private. Configure it in your client’s authentication settings rather than including it in prompts or committing it to version control.
For example, ask your assistant:
Find 10 current engineering jobs in Amsterdam. Look up the department and location filter values first, then show the job titles, companies, and links.
Or, for a specific company:
Find 10 current jobs at Google using the company LinkedIn URL https://www.linkedin.com/company/google/.
When building searches:
  • Company identity: Both company-search tools require company_linkedin_url.
  • Filters: Use department and job board values exactly as returned by the lookup tools. For locations, pass the returned value objects in jobs_locations or exclude_jobs_locations, rather than display labels.
  • Page size: Request 10–100 jobs per page to keep results manageable.
  • Pagination: Pass the response’s search_after cursor unchanged into the next call, keeping the same filters. Stop when no next cursor is returned.
  • History: Every plan includes historical searches covering postings back to 2020. Specify the date range you want your assistant to search.

Credits and authentication

Searches consume your account’s API credits. Department, job board, location, and credit-balance lookups consume no credits and work with zero or negative balances. Each HTTP request is authenticated. With OAuth, the server checks your sign-in with Sentrion’s identity provider and looks up your account’s API key. With an API key, it validates the key through the credit-balance endpoint. Either way, tool calls use your account’s key for that request, so results and billing belong to your account. Revoked or inactive keys cannot connect, and authentication fails if a validation service is unavailable.

Python example

With the fastmcp package installed, set SENTRION_API_KEY in your environment and run this example to list the tools and check your balance:

Troubleshooting

API HTTP errors surface as MCP tool errors. Also inspect the search response’s success and error fields: the API can return a search failure in an HTTP 200 response. See response formats and errors.