القائمة
Quality-check repository GitHub issues and issue comments. Use when creating, updating, reviewing, or validating issue titles, bodies, replies, or thread notes.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Review or create a BepInEx Mono mod for C# project and module structure, plugin identity, game compatibility, dependency boundaries, repository-family alignment, GitHub CI and release automation, package readiness, release metadata, and verification. Use when implementing, reviewing, preparing a release for, diagnosing quality gaps in, or applying a minimal-difference peer rollout to a BepInEx 5 Mono plugin or its build and release workflow.
Set up and use an isolated Git worktree for repository implementation tasks. Use unless the user explicitly requests another workspace arrangement.
Quality-check repository pull requests and PR-thread communication. Use when creating, updating, reviewing, or validating PR titles, bodies, review comments, replies, thread notes, or GitHub squash-merge attribution and commit-message payloads.
Review Agent Skills for trigger clarity, focused scope, metadata alignment, progressive disclosure, and validation readiness. Use when creating, updating, reviewing, or splitting a skill's SKILL.md, references, scripts, assets, or agents/openai.yaml metadata.
Investigate version-specific Lethal Company mechanics using verified Unity-build evidence, seeded RNG, exhaustive or sampled computation, and runtime observation; produce reproducible, bounded gameplay reports. Use for mechanics questions, full-seed enumeration, sealed-result reuse, Terminal or HUD endpoint interpretation, version comparisons, statistical prediction, or playable-strategy evaluation. Use unity-game-analyze for the underlying decompiled-code and serialized-asset trace.
Roll out a validated canonical change from a designated primary repository to compatible peers with only necessary local differences. Use for shared Agent Skills, repository rules, workflows, or documentation conventions.
| name | issue-quality-check |
| description | Quality-check repository GitHub issues and issue comments. Use when creating, updating, reviewing, or validating issue titles, bodies, replies, or thread notes. |
Identify whether the artifact is an issue title, body, reply, or combined update.
For AI-agent-prepared issue text, add the applicable alert at the very top:
> [!WARNING]
> This issue was created with assistance from LLMs.
Use this comment alert instead for replies:
> [!WARNING]
> This comment was created with assistance from LLMs.
Omit an alert only when assistance was not significant.
Make titles concise and specific to the problem or task. Include an affected area when useful; avoid vague titles and do not impose Conventional Commits unless repository policy requires it.
Keep the body or reply concise. Use only useful sections: Summary, Details, Acceptance Criteria,
Verification, Notes, Findings, or Next Steps. For bugs, state expected and actual behavior and known useful
reproduction steps. Cross-link durable follow-up origin context when it helps triage.
Write in English except for exact source material. Use bullets, backticks for technical names, explicit uncertainty and verification limits, and summaries rather than large logs or diffs. Never describe AI-performed work as manual.
Use prose-quality-check for nuanced explanatory prose and security-check for security- or supply-chain-sensitive
content.
Add Update Note or Discussion Note only when the task requests process or decision notes. Place
Request addressed: ... after the required alert and before the note heading; label inferences, avoid secrets and
private paths, and keep normal replies direct.
When using gh, write Markdown to a temporary file and pass --body-file; use confirmed identifiers or placeholders
such as <issue-number>. Verify stored issue bodies with gh issue view --json body and stored replies when possible,
correct quoting, then remove temporary files.