Connect AI agents to CronLite's MCP cron scheduler

Connect over Streamable HTTP

Every CronLite instance embeds an MCP endpoint at /mcp. Configure your client's Streamable HTTP connection with the server URL and an Authorization bearer token. Use HTTPS when connecting over a network and keep the API key in your client's secret storage.

Endpoint: http://localhost:8080/mcp
Transport: Streamable HTTP
Authorization: Bearer ec_<your-api-key>

Use the stdio proxy with Claude Desktop

For clients that require stdio, build cronlite-mcp and point the client at that executable. CRONLITE_URL selects the running CronLite server; CRONLITE_API_KEY provides the namespace key. This example uses Claude Desktop's MCP configuration format.

go build -o cronlite-mcp ./cmd/cronlite-mcp

{
  "mcpServers": {
    "cronlite": {
      "command": "/absolute/path/to/cronlite-mcp",
      "env": {
        "CRONLITE_URL": "http://localhost:8080",
        "CRONLITE_API_KEY": "ec_<your-api-key>"
      }
    }
  }
}

Ten tools for the job lifecycle

Use resolve-schedule to turn a supported description such as every weekday at 9am into a cron expression. Review the resolved expression and timezone before calling create-job. Supply name, cron_expression, timezone, and webhook_url; add a webhook_secret for receiver verification.

• create-job, list-jobs, get-job, update-job, delete-job
• pause-job, resume-job, trigger-job
• next-run, resolve-schedule

Keep agent access scoped and observable

Each MCP request is authenticated with a namespace-scoped API key. The key limits which jobs and executions the client can access. Treat job deletion and immediate triggering as consequential operations in your agent workflow.

Use next-run to preview timing and get-job to inspect recent executions. The Admin UI offers a visual companion for reviewing the jobs your agent creates. Webhook retries and receiver deduplication still apply regardless of whether a job was created through REST, Admin, or MCP.