Building in Claude/ChatGPT via MCP

Building automations from an external AI client

If you use Claude Desktop, Claude Code, or another MCP-compatible AI client, you can connect it to your Chase Agents workspace and build automations directly from within that tool. Your external client gets access to the full Chase Agents execution engine via the MCP server.

The build workflow

1. Call list_connections to see all APIs and MCP servers available in your workspace
2. Read connection details with read_resource to understand authentication and available endpoints
3. Test the workflow with execute_steps, verifying each step output before proceeding
4. Save the tested steps as an automation with create_automation
5. Schedule it with manage_schedules if needed

Why build via an external MCP client

Building via an external client like Claude gives you that client's full reasoning and coding capabilities combined with Chase Agents' full execution and persistence capabilities. Complex automations requiring careful data transformation, error handling, and multi-step coordination are often easier to build in a long conversational session with a capable model.

Accessing guide documentation

The MCP server exposes detailed guide resources. From any connected external client, call read_resource with type guide and name automation-building for the full automation building guide, or type guide and name step-types for the complete step type reference.