メニュー
Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Ultra-compressed communication mode. Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping full technical accuracy. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman-thayto.
Sets up an `## Agent skills` block in AGENTS.md/CLAUDE.md and `docs/agents/` so the engineering skills know this repo's issue tracker (GitHub or local markdown), triage label vocabulary, and domain doc layout. Run before first use of `to-issues-thayto`, `to-prd-thayto`, `triage-thayto`, `diagnose-thayto`, `tdd-thayto`, `improve-codebase-architecture-thayto`, or `zoom-out-thayto` — or if those skills appear to be missing context about the issue tracker, triage labels, or domain docs.
Use whenever the user wants to review and simplify the code they just wrote on the current branch — refactor with KISS and DRY, kill duplicates, strip unnecessary comments, then fix TypeScript errors and warnings, then verify tests pass. Triggers on `/simplify-thayto`, and on phrases like "simplify our changes", "simplify the code we just wrote", "review and improve our changes", "DRY this up", "refactor with KISS", "deduplicate this", "tighten up the diff", "clean up what we did", "kill the duplication", "remove unnecessary comments", "strip dead comments", "kill the comment noise", or any equivalent ask to revisit and tighten the branch diff before shipping. Applies hard structural rules — avoid classes, no nested ifs, no nested ternaries, prefer small functions — keeps only WHY comments and removes WHAT-comments / tombstones / ticket references, and delegates type-level work to the `typescript-advanced-types` skill so types stay simple. Never creates a branch and never commits; edits the working tree in pl
Use whenever the user wants to self-review their own pull request before flagging it for human review — runs the same multi-agent review (security, performance, TypeScript, code-quality) as `pr-review-opinionated-thayto`, then applies the `address-my-prs-thayto` rubric (apply / escalate / skip) to act on each finding locally without posting anything to GitHub. Triggers on `/self-review-thayto`, "self-review", "self review my PR", "review my own PR before sending it out", "pre-flight my PR", "review and address my own PR", "fix my PR before review", "review my PR myself first", or pasting the user's own PR URL with phrasing like "review this and fix it" or "go through this myself first". Operates on a single PR — the current branch's PR (resolved via `gh pr view --json number`) or a pasted github.com/<owner>/<repo>/pull/<n> URL. NOT for reviewing someone else's PR (use pr-review-opinionated-thayto). NOT for batch-triaging multiple PRs (use address-my-prs-thayto).
Use whenever the user wants to triage, babysit, or address feedback across all of their open pull requests on the current GitHub repo at once. Triggers on phrases like "check my PRs", "address PR comments on my PRs", "fix CI on my PRs", "triage my open PRs", "babysit my PRs", "go through my pull requests", or pasting a PR-list URL like https://github.com/<owner>/<repo>/pulls (note: list URL, not a single /pull/<n>). Triggers when the user references multiple of their own PRs (author resolved dynamically from `gh api user` and filtered with `--author @me` — never hardcoded; e.g. rafa-thayto when that's the authenticated login) and wants review comments addressed, broken CI fixed, base-branch rebases performed, threads resolved silently (the skill never replies on a thread — only resolves or escalates), and reviewers re-requested. NOT for reviewing a single PR as a reviewer (use pr-review-opinionated-thayto). NOT for opening a new PR.
Use this skill whenever the user wants to scaffold a new issue, ticket, todo, or task file — phrases like "create an issue", "create a ticket for X", "add an issue", "new todo", "open a ticket", "file an issue", "create a linear issue", "linear ticket", "write up a task", or any equivalent. Creates a numbered markdown file under `docs/issues/` with YAML frontmatter (id, status, title) and the standard sections Description / Why / Acceptance / What you must do / Hints. Auto-increments the id by scanning existing files. Conducts a brief consolidated interview only when the user's request is too thin to write a useful ticket — otherwise scaffolds directly. Trigger this skill even when the user says "linear" or talks about a tracking system; the skill writes a local file regardless of which tracker they mention.
| name | zoom-out-thayto |
| description | Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture. |
| disable-model-invocation | true |
I don't know this area of code well. Go up a layer of abstraction. Give me a map of all the relevant modules and callers, using the project's domain glossary vocabulary.
