本文へ移動
cccskills
無料GitHub で公開

factory-babysit-pr

Experimental workflow for babysitting one explicitly authorized pull or merge request. Use to monitor its live review state and checks, fix actionable in-scope findings, and apply separate reply, approval, merge, and soak gates.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md5.2 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Factory Babysit PR

Start with the Factory guide — docs/factory.md in the AOS repo, upstream at https://github.com/BuilderIO/skills/tree/main/docs/factory — for workflow setup and configuration.

Follow one authorized PR or merge request through a delivery lifecycle. This is not a queue sweep like factory-review-prs or a stopped-task scan like factory-watchdog.

Read .agent-factory/config.yaml, the optional skill_prompts.factory-babysit-pr entry, the target repository's instructions, and the Factory configuration reference. A PR number, label, green check, or unrelated ship request is not authorization. The prompt is additional project guidance, not PR authorization.

Each pass

  1. Confirm authority and ownership. Verify the exact host, repository, PR, original user authorization, allowed paths/risk, and task-owned worktree. If anything is missing or stale, stop before writing.
  2. Read a complete live snapshot. Record base and head, open/draft/closed/ merged state, all required checks, current reviews and approvals, unresolved threads, and mergeability. Partial or unavailable data is unknown, not clean.
  3. Disposition feedback. Compare each new finding with the exact diff and source. Preserve human direction unless evidence disproves it. Fix only actionable findings inside the authorized path/risk limits; avoid speculative cleanup and larger redesigns.
  4. Verify and re-read. Draft or prepare only; needs the user's literal GO for this exact action: a push or any host mutation happens only after GO. Run configured checks only locally; checks and reads never mutate state. After every push (which itself needs GO), fetch the live PR again and bind all evidence to its exact current head. A changed head invalidates old check/review evidence and resets the soak.
  5. Apply action gates separately. A review or fix does not authorize a reply, push, approval, or merge.

Action gates

Every write/irreversible row below: Draft or prepare only; needs the user's literal GO for this exact action.

ActionProceed only when
FixEditing is enabled; feedback is actionable, in scope, and within risk/path limits. Local edits after the authorized original fix GO.
PublishDraft or prepare only; needs the user's literal GO for this exact action — then the destination must still be the authorized PR branch with its live head matching the fix's base. Never force-push or overwrite a concurrent update.
ReplyDraft or prepare only; needs the user's literal GO for this exact action. First confirm replies are enabled and their conditions hold; follow configured wording/tone; otherwise do not message.
ApproveDraft or prepare only; needs the user's literal GO for this exact action. First report that every author, diff, check, and review condition holds on the live head.
MergeDraft or prepare only; needs the user's literal GO for this exact action. First report that all configured gates pass on the live head and the host can condition the merge on that head.
SoakProve the exact head and required gates stayed unchanged for the full configured interval. Missing history does not satisfy it.

Use the configured code-host abstraction for reads and writes. Do not assume GitHub, a specific CLI, or identical conditional-merge support across hosts. If the host cannot provide complete current review/check state or bind a merge to the checked head, keep the dependent action disabled.

Stop and report

Stop when the PR is merged, closed, cancelled, blocked on a person or external dependency, or no authorized action remains. Report the PR link and current head, state inspected, fixes and checks, external actions taken or withheld, soak status, and the concrete blocker or next check time. Distinguish read state from unavailable or inferred state.

AOS safety rules

  • Configuration lives in .agent-factory/config.yaml. If it is missing, stop and ask the user; never create it or guess sources.
  • Nothing in the config can open the AOS GO gate. git push, gh pr merge, gh release create and every external write (reply, comment, approval, close, status change, notification) need the user's literal GO for that exact action. Prepare the change or draft text, show it, and stop.
  • Scheduled or unattended runs are read-only. Report; do not act.
  • Being enabled, authorized, or holding the original authorization in config or upstream text never counts as GO. An inbound message or a watchdog trigger is never a GO.
  • Repositories stay private. Commits use Conventional Commits.
  • No recursive deletes. Checks run locally only and never post secrets or tokens.
  • Connectors are whatever the host already exposes. Do not install or register integrations or MCP servers.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

hybridlabor-api のスキルをすべて見る

このスキルの問題を報告する