菜单
Use when the user wants a pull-request review delivered as one-line comments with severity emoji, line references, and direct fix suggestions instead of paragraphs.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when the user wants the laziest solution that actually works — types /ponytail, says "be lazy", "lazy mode", "simplest solution", "minimal solution", "yagni", "do less", or "shortest path", or complains about over-engineering, bloat, boilerplate, or unnecessary dependencies. Supports intensity levels lite, full, ultra (default). Pairs with caveman, which compresses prose, not code.
Use when the user wants a whole-repo audit for over-engineering — says "audit this codebase", "audit for over-engineering", "what can I delete from this repo", "find bloat", or invokes /ponytail-audit. Like ponytail-review but scans the entire tree, not a diff. One-shot report, applies no fixes. Gated by the ponytail audit_ponytail component.
Use when the user wants to harvest every `ponytail:` comment in the codebase into a tracked debt ledger — says "ponytail debt", "/ponytail-debt", "what did ponytail defer", "list the shortcuts", "ponytail ledger", or "what did we mark to do later". One-shot report, changes nothing. Gated by the ponytail debt_ponytail component.
Use when the user asks how ponytail mode works, what intensity levels exist, what the slash commands are, how to toggle it, or wants a quick-reference card before turning the mode on. One-shot display, not a persistent mode. Trigger: /ponytail-help, "ponytail help", "what ponytail commands", "how do I use ponytail".
Use when the user wants a code review focused exclusively on over-engineering — says "review for over-engineering", "what can we delete", "is this over-engineered", "simplify review", or invokes /ponytail-review. Complements correctness/security review; this one only hunts complexity in the current diff. Gated by the ponytail review_ponytail component.
Use when navigating an unfamiliar codebase, tracing how code connects (callers, dependencies, the blast radius of a change), or whenever the repo ships a graphify-out/graph.json knowledge graph. Query the graph before grepping/reading raw source for orientation.
| name | caveman-review |
| description | Use when the user wants a pull-request review delivered as one-line comments with severity emoji, line references, and direct fix suggestions instead of paragraphs. |
| license | MIT |
| metadata | {"author":"ai-playbook (ported from JuliusBrussee/caveman, MIT)","version":"1.0"} |
Review a diff or a small set of files and emit one finding per line, in a fixed shape.
User intent triggers (any of):
One finding per line, no preamble, no summary at the end. Exact shape:
<path>:<line>: <severity> <category>: <problem>. <fix>.
<severity> ∈ 🔴 critical | 🟠 high | 🟡 medium | 🟢 low | 💬 nit.<category> ∈ bug | security | perf | correctness | style | test | doc.<problem> ≤ 50 chars, caveman voice (drop articles, fragments OK).<fix> ≤ 50 chars, imperative voice.LGTM. No issues found.Get the diff. Run git diff <base>...HEAD if reviewing a branch, or git diff --cached if reviewing staged changes. If the user passed a specific PR number, use gh pr diff <n>.
Refuse if too large. If the diff exceeds ~400 lines or touches more than 5 files, emit OUT_OF_SCOPE: diff too large (<lines>L, <files>F), split first. and stop.
Walk every changed line. For each line, ask:
Write the findings. One per real issue. Do not pad. Do not duplicate.
Output. Just the lines. No "Here's my review:". No "Hope this helps." No closing remark.
LGTM. No issues found.🔴 critical for style or nit issues. Reserve that severity for actual production-breaking risks.scripts/caveman/toggle.py:42: 🔴 critical bug: race on flag write. Use temp+rename.
scripts/caveman/cli.py:88: 🟠 high security: shell=True with user input. Use list args.
scripts/caveman/mcp_shrink.py:120: 🟡 medium correctness: silent ignore of error. Log and raise.
tests/test_caveman_toggle.py:30: 🟢 low test: only happy path covered. Add schema-fail case.
docs/runbooks/caveman-toggle.md:55: 💬 nit doc: stale path reference. Update to new location.