Resume a previously started 오토파일럿 project from saved state files. Use when the user asks to continue, resume, keep going, or finish an existing autopilot project.
Installs into .claude/skills of the current project.
Are you the author of Autopilot Resume?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/hashgraph-online-autopilot-resume)
---
name: autopilot-resume
description: Resume a previously started 오토파일럿 project from saved state files. Use when the user asks to continue, resume, keep going, or finish an existing autopilot project.
summary: Read state, pick the next task, continue execution.
metadata:
priority: 7
promptSignals:
phrases:
- "오토파일럿 계속"
- "오토파일럿 재개"
- "resume autopilot"
- "continue autopilot"
- "keep going"
- "다음 작업 이어서"
anyOf:
- "resume"
- "continue"
- "재개"
- "계속"
minScore: 4
retrieval:
aliases:
- 오토파일럿 재개
- resume autopilot
- continue execution
intents:
- continue a saved project
- resume from progress files
entities:
- spec
- progress
- next
- blockers
---
# Auto Pilot Resume
Use this skill when a project already has saved autopilot state.
## Required Inputs
Read these files first if they exist:
- `autopilot/state.json`
- `autopilot/blockers.json`
- `autopilot/secrets-status.json`
- `docs/next.md`
- `docs/spec.md` when setup is already complete
- `docs/design.md` for user-facing projects when setup is already complete
- `docs/progress.md` when setup is already complete
## Resume Workflow
1. Read `executionMode`, `teamRoles`, `currentOwner`, `qualityGates`, `lastPlannerCheckpoint`, `lastReviewerVerdict`, and `setupStatus`.
2. If `setupStatus` is `pending` or `autopilot/secrets-status.json.status` is `pending`, treat resume as setup continuation.
3. In that setup continuation path:
- surface only the missing env payload
- treat the project as a valid existing project even if `docs/spec.md` and `docs/progress.md` do not exist yet
- do not resume implementation work until setup is complete
4. Otherwise determine whether there is an active blocker.
5. If the blocker is `human-required`, surface only the next minimal action.
6. If not blocked, confirm whether native sub-agents are currently available.
7. If the saved mode assumes native specialists but they are unavailable, downgrade the runtime path to `serial-fallback` and update state/progress/next before continuing.
8. Resume from the current owner role and the highest-priority unfinished task.
9. Preserve the manager flow:
- planner before builder
- designer review for user-facing UI slices
- QA verdict before completion
10. Implement, validate, and update the state files.
- use `python3 ../../scripts/team_checkpoint.py ...` after each planner / architect / builder / designer / QA result so resume state actually advances
11. Stop only if:
- definition of done is met, or
- a human-required blocker prevents further work
## Resume Output
Return a concise status update:
- last completed work
- current owner role
- current task
- next task after this
- QA verdict
- blocker or setup status