---
name: mission-control-debug
description: Use when Mission Control orchestration is stuck, failed, degraded, or unclear from Codex chat.
---

# Mission Control Debug

## Purpose

Explain why Mission Control is stuck and surface the safest recovery path through Codex chat.

## Bridge Rule

The Codex chat agent is not the Mission Control Manager. It is the bridge.

## When To Use

- The run looks stuck
- The daemon or bridge is unhealthy
- The user asks why orchestration failed

## Required Mission Control Tools, Resources, And Prompts

- Tools: `mission_control_get_status`, `mission_control_get_diagnostics`, `mission_control_get_pending_decisions`, `mission_control_get_orchestration_events`, `mission_control_request_recovery_options`
- Resources: `mission-control://projects/{project_id}/diagnostics`, `mission-control://orchestrations/{orchestration_id}/events`, `mission-control://projects/{project_id}/pending-decisions`
- Prompts: `debug-failed-orchestration`, `continue-orchestration`

## Step-By-Step Workflow

1. Fetch status.
2. Fetch diagnostics.
3. Fetch pending decisions.
4. Fetch recent orchestration events.
5. Ask Mission Control for recovery options if the tool is available.
6. Return a concise blocker explanation.

## What To Show The User In Codex Chat

- Current failure or degraded state
- Whether the run is blocked on the user, infrastructure, or manager recovery
- Relevant recent events
- Safe next options

## Safety And Approval Behavior

- Keep raw logs, secrets, and stack traces out of the default summary.
- Treat retries or resumes as explicit actions, not assumptions.

## Fallback Behavior If The Daemon Is Unavailable

- State that bridge-level diagnostics could not reach Mission Control.
- Suggest verifying the daemon first before continuing.

## What Not To Do

- Do not claim a recovery worked without rechecking status.
- Do not confuse waiting for approval with a crash.
- Do not invent recovery options that Mission Control did not return.