Skip to content

MCP control surface

MCP is a control surface for Orcho. It lets a client start, observe, diagnose, and decide runs without scraping terminal output.

If you have not configured MCP yet, start with Start with MCP. If you are deciding whether MCP is worth the extra orchestration cost, read LLM captain mode. If you are building a client-side lifecycle flow, read Expert MCP control loop.

Use MCP when you want a client to:

  • start a run;
  • poll status;
  • read evidence slices;
  • diagnose why a run stopped;
  • decide a phase handoff;
  • ask an LLM advisor for a handoff recommendation;
  • decide a delivery gate;
  • resume or recover from a typed next action.
ToolPurpose
orcho_run_startSpawn a detached run.
orcho_run_statusRead the current run snapshot.
orcho_run_events_tailFollow durable event progress.
orcho_run_evidenceRead typed evidence slices.
orcho_run_diagnoseAsk what the next safe action is.
orcho_run_resumeRe-spawn a run when resume is meaningful.
orcho_handoff_adviceAsk for a read-only recommendation on a paused handoff.
orcho_delivery_gateInspect post-release delivery or correction state.
orcho_delivery_decideResolve a parked delivery decision.

MCP makes Orcho more than a CLI transcript. It gives agent clients a stable protocol for lifecycle control and evidence inspection.

The client should not need to guess whether a run is done, blocked, or waiting for input. It should read that from the run-control surface.

The best division of labor is simple: use the CLI when the live stream is the main value; use MCP when the client should keep state, diagnose stops, and help the operator choose the next action.