longrun-resume

star 20

Resume the latest or a specified long-running Copilot CLI mission from .copilot-mission-control/ without restarting completed work. Use when the user asks to continue, resume, or cleanly converge a previous /longrun run.

izscc By izscc schedule Updated 4/5/2026

name: longrun-resume description: Resume the latest or a specified long-running Copilot CLI mission from .copilot-mission-control/ without restarting completed work. Use when the user asks to continue, resume, or cleanly converge a previous /longrun run. allowed-tools: "*" user-invocable: true disable-model-invocation: false

用它继续一个已有 run,或把脏状态 run 收敛到 COMPLETE / BLOCKED

1. Resolve run

  • 若 prompt 指定 run-id,就用它。
  • 否则读取 .copilot-mission-control/state/latest-run-id
  • 若不存在 run,明确提示先用 /longrun <任务描述>

2. Restore context

先读:

  • mission.md
  • plan.md
  • status.json
  • journal.jsonl
  • COMPLETION.mdfinal-summary.md(若存在)
  • hook-events.jsonl 尾部(仅在需要解释错误时)

3. Resume rules

  • status.state 已是 complete,默认只读,不重跑。
  • status.state 已是 blocked,先解释阻塞原因,只在用户明确要求 reopen 时才继续。
  • 若 deliverable 已存在、COMPLETION.md / final-summary.md 缺失或 status.json 仍是 running,优先做本地验证 + finalize 收敛,不要重新做已完成工作。
  • 只有在产物未完成时,才继续未完成 workstreams。

4. 恢复策略

  • 设置 .copilot-mission-control/state/active-run-id 为目标 run。
  • 恢复 status.json.state=running 仅在确需继续执行时进行。
  • 优先沿用既有 profilelanguagemodelPolicydeliverablescompletedWorkstreams
  • 使用最小恢复路径,不重复已完成 workstreams,除非验证表明它们失效。

5. finalize 优先

若本地验证已足以证明任务完成,直接调用:

python3 "$HOME/.copilot-mission-control/bin/finalize_run.py" \
  --workspace "$PWD" \
  --run-id "<run-id>" \
  --status complete \
  --headline "Resumed run converged via local verification" \
  --local-verify

若恢复预算耗尽或缺依赖,finalize 为 blocked,不要无限空转。

Install via CLI
npx skills add https://github.com/izscc/LongRun --skill longrun-resume
Repository Details
star Stars 20
call_split Forks 4
navigation Branch main
article Path SKILL.md
More from Creator