SKILL.md · Claude Code, Cursor, Codex
skill-review
description
Use when auditing, tightening, or deciding whether to keep one agent SKILL.md. Not for writing product code or inventing skill content you have not read.
- What it does
- A 0–2 rubric your agent can use to audit and tighten any SKILL.md, including your own.
- Where it stops
- Not for writing product code or inventing skill content you have not read.
One-time payment. The zip has all 11 skill folders with their references.
- ---
- name: skill-review
- description: Use when auditing, tightening, or deciding whether to keep one agent SKILL.md. Not for writing product code or inventing skill content you have not read.
- ---
- # Skill review
- Audit one `SKILL.md` (and optionally tighten it) against a five-line rubric for skill craft, routing, and catalog fit.
- **Scoring detail:** [references/rubric.md](references/rubric.md)
- ## How
- Read the target frontmatter and body, plus its `references/` and `scripts/` if present. Score it with the rubric, including a quick sibling scan for when-to-load collisions with other skills in the same library. Prefer regenerating a weak section from its goal and constraints over stacking patches. Leave runtime discovery (which files exist, what the error says) to the agent that will run the skill under review. Do not hard-code their paths here.
The other 10 skills in the pack
- technical-seo-2026-delta
Use when a site needs only the 2025–2026 Google/Bing technical changes checked (generative eligibility, snippet controls, dead rich results, crawler split, new Search Console reports and measurement caveats, EEA site-reputation split, regional SERP units, emergency crawl-rate controls).
- aeo-answer-surface-playbook
Use when the goal is to own the answer unit on a named results/answer surface (featured snippet, PAA, AI Overview, Copilot, Perplexity), especially when the site already ranks and still loses the box.
- geo-citation-playbook
Use when a brand needs to be retrieved and cited inside generative answers (ChatGPT Search, Copilot, Perplexity, Claude, Gemini, Google AI Mode) — per-engine eligibility, citation measurement, third-party corroboration.
- programmatic-seo-gate
Use when someone wants template, location, integration, comparison, directory, or other pages at scale (programmatic SEO) and needs a 2026 allow/refuse decision with a page cap.
- site-ia-map
Use for a hub/spoke, cannibalization, one-URL-per-intent, internal-links, or site information-architecture job (live remap or greenfield).
- search-change-triage
Use every Monday, or when traffic / rankings / AI Overviews / AI citations moved, to classify the external search change and then choose wait, watch, or act.
- gsc-growth-review
Use when the ask is a Google Search Console period review and a qualitative SEO growth judgment for a named property.
- seo-agent-design
Use when someone asks you to build an SEO / AEO / GEO agent or content bot and you need its safety contract (tools, approve path, refuse list) before any code.
- how-to-site-factory
Use when building or rebuilding a vertical how-to / niche publisher site with a coding agent (repo, stack defaults, SEO defaults, image, accuracy and byline rules).
- linkedin-post-system
Use when designing a repeatable LinkedIn publishing system (topics, formats, cadence, review) that stays inside LinkedIn's 2026 rules, e.g. for building in public.