AI Agent Hub
Back to skills
Mission Control icon

Mission Control

AI Agent Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md to install @user_b26749d2/3c12646.

About this skill

What Problem It Solves

Before an agent edits code or files, common risks are misaligned plans, missing execution traces, and weak post-mortem. Mission Control turns tasks into a trackable workflow so the agent confirms goals, constraints, acceptance criteria, and risks before acting.

How It Works

It applies a plan-confirmation model for code or file operations:
- Generate t-requirement.md and t-plan.md to capture task requirements and execution plan.
- Complete 3 truth-seeking, analysis, and revision iterations in the planning stage; use qiushi for critique-style review when needed.
- Execute only after Boss confirmation, with the main agent Moss recording progress in t-log.md.
- Close with t-report.md, preserving task ID, acceptance criteria, risk plan, and version changes.

It can also convert a static specification into a task card: keep the original scientific content, data, and constraints, then fill in task ID, output requirements, and Moss fields without overwriting the source by default.

Scope and Caveats

Use it for tasks that need human confirmation, traceability, and review. Avoid it for casual chat, pure lookups, or small requests with no file changes. If expected completion time is missing, mark it as pending. Do not treat a static specification as a ready-to-run task card without conversion.

Use Cases

  • Before a code refactor, produce requirement, plan, log, and report, then execute only after Boss approval.
  • Convert a static project spec into a task card, adding acceptance criteria, output requirements, and Moss fields.
  • When a task errors or pauses, require the log to state the cause and produce a closing report for Boss.
  • Run three truth-seeking, analysis, and revision iterations in planning before execution is allowed.

Best For

  • Backend engineers who need agents to confirm plans and leave traces before code or file edits
  • Research assistants who need to convert project specs into task cards with acceptance criteria
  • Test leads who require agents to log failure causes and produce closing reports
  • Product managers who need repeated plan iterations before approving execution