Skip to content

MCP setup

This is the shortest path to a working PlanExe MCP integration.

Recommended: point your MCP client at the hosted server https://mcp.planexe.org/mcp using a pex_... key from home.planexe.org. Only fall back to local Docker (http://localhost:8001/mcp) if you need to run the stack yourself. Grab the JSON/CLI snippet from MCP client config snippets or the per-client guide below.

1. Understand the flow

  1. Ask for prompt examples.
  2. Inspect model_profile options and available models.
  3. Expand the user idea into a high-quality prompt (typically ~300-800 words) and get user approval. Use this compact shape: objective, scope, constraints, timeline, stakeholders, budget/resources, and success criteria.
  4. Create the plan.
  5. Poll for status (about every 5 minutes).
  6. If status is failed, optionally call plan_retry (defaults to model_profile=baseline).
  7. Download artifacts via plan_file_info.

2. Minimal tool usage

  1. example_plans (optional, preview example output)
  2. example_prompts
  3. model_profiles
  4. plan_create
  5. plan_status
  6. plan_retry (optional, only for failed plans)
  7. plan_file_info

For plan_create:

3. Success criteria

  • You can fetch example prompts.
  • You can create a plan.
  • You can fetch artifact metadata/URLs with plan_file_info.
  • Your client can parse error.code and error.message and handle {} from plan_file_info as "not ready yet".
  • If running parallel work, your client tracks multiple plan_ids explicitly (server-side global cap is not enforced).

Next steps