- Create a Checkr account (or use an existing one) in the target environment (Production or Staging).
- Go to Account → Developer Settings and click Generate MCP Credentials.
You will receive:
- Client ID
- Client Secret
- Token Endpoint
These are unique to your Checkr account. Keep them secure — you'll need all three in the next step. The Client Secret is shown only once; copy it immediately.
Establish a secure connection by fetching an OAuth access token. This allows your system to communicate with Checkr's APIs securely.
- Flow: Client Credentials
- Action: Exchange your Client ID and Client Secret for a bearer access token at the Token Endpoint provided during credential provisioning.
- Security: Store credentials in protected environment variables — never hard-code them.
POST <token_endpoint>
Content-Type: application/x-www-form-urlencoded
grant_type=client_credentials&client_id=<ID>&client_secret=<SECRET>Point your MCP client at the Checkr server using the token from the previous step.
| Setting | Value |
|---|---|
| Endpoint | https://mcp.checkr.com/candidate/ (Production) or https://mcp.checkr-staging.com/candidate/ (Staging) |
| Auth | Authorization: Bearer <token> on every request |
| Transport | Streamable HTTP (session-stateful — each connection maintains identity state) |
Once connected, the MCP protocol handles the rest — your LLM automatically discovers the available tools, their parameters, and descriptions directly from the server. No manual tool configuration is needed.
After the initial connection, the server returns an Mcp-Session-Id header. This header must be included on all subsequent requests within the same session — without it, each request creates a new session and any candidate binding is lost. If you're using an agent framework with built-in MCP support (e.g., Claude, OpenAI Agents SDK, LangChain), session persistence is handled automatically. If your agent uses a custom HTTP client to connect to the MCP, ensure it preserves and forwards this header across requests.