What does this skill do, and when should you use it?
geo-audit is one of 16 skills bundled in the zubair-trabzada/geo-seo-claude repository (at skills/geo-audit/SKILL.md) and serves as the orchestration core for full Generative Engine Optimization (GEO) audits. It fetches a site's homepage and sitemap, auto-detects the business type (SaaS, local business, e-commerce, publisher, agency), then delegates parallel analysis to 5 subagents covering AI visibility, platform readiness, technical infrastructure, content E-E-A-T, and schema markup. Category scores are combined into a weighted composite GEO Score (0-100), and the skill writes a GEO-AUDIT-REPORT.md with severity-ranked issues and a 30-day action plan. The MIT-licensed repo ships the collection via install.sh; note that sibling capabilities like PDF reports and brand scanning belong to other skills, though this skill delegates to corresponding subagents during the audit.
1) Fetches the homepage via WebFetch, extracting title, navigation, footer, and schema signals to classify the business type; 2) Reads sitemap.xml or crawls internal links (max 50 pages, 30-second per-page timeout, robots.txt respected, minimum 1-second delay between fetches); 3) Delegates to 5 parallel subagents scoring: AI visibility (citability, AI crawler access, brand authority), platform optimization (Google AI Overviews/ChatGPT/Perplexity/Gemini/Bing Copilot), technical GEO, content E-E-A-T, and schema/structured data; 4) Computes the composite GEO Score with weights (citability 25%, brand authority 20%, E-E-A-T 20%, technical 15%, schema 10%, platform 10%); 5) Classifies all issues as Critical/High/Medium/Low and writes GEO-AUDIT-REPORT.md including category deep dives, weekly quick wins, and a 30-day action plan.
- A GEO/SEO agency needs a client-ready website AI-visibility audit with scores and a prioritized action plan.
- A SaaS company wants to know why its product doesn't appear in ChatGPT or Perplexity answers.
- A local business owner wants to be recommended by AI assistants and needs LocalBusiness schema and local signals checked.
- A content team wants to assess how citable a newly published article is (citability scoring).
- An e-commerce operator wants to verify product schema, review aggregation, and buying guides meet AI shopping recommendation requirements.
- A marketing team inheriting a new site wants a prioritized technical and content issue list before restructuring.
How do you install this skill?
- This is a static, non-executed review; audit workflow and scoring quality are unverified at runtime.
- Core functionality depends on WebFetch reaching overseas sites and platforms (YouTube, Reddit, Wikipedia); likely degraded or unusable from mainland-China networks, and no Chinese-language support is declared.
- Bash and Write are granted; review what commands may run at runtime before installing.
- Marketing statistics in the description and README (+527%, 4.4x, 3x, etc.) lack citations and should not drive decisions.
- The curl|bash install path executes remote code; prefer cloning and inspecting install.sh before running.
- Shell / CLI
- Network access
- Local filesystem
Claude Code CLIPython 3.8+Git
Install the whole skill collection (this skill ships with the repo): on macOS/Linux run curl -fsSL https://raw.githubusercontent.com/zubair-trabzada/geo-seo-claude/main/install.sh | bash, or manually git clone https://github.com/zubair-trabzada/geo-seo-claude.git && cd geo-seo-claude && ./install.sh; on Windows use install-win.sh from Git Bash. Requirements: Claude Code CLI, Python 3.8+, Git (plus python3-venv on Debian/Ubuntu; optional uv for faster installs). Python dependencies go into an isolated venv at ~/.claude/skills/geo/.venv/ without touching system Python. The repo does not document installing only the skills/geo-audit skill in isolation.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/zubair-trabzada/geo-seo-claude.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/geo-audit" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
In Claude Code, run an audit against any website — the README's entry point is /geo audit <url>, routed by the main geo skill to this orchestration skill. The skill automatically fetches, classifies, runs parallel analysis, and produces GEO-AUDIT-REPORT.md. Inspect the report and intermediate data with Read/Grep. Note that SKILL.md itself defines no standalone slash command; the actual trigger depends on routing in the parent geo/SKILL.md.
What are this skill's strengths and limitations?
- Complete audit pipeline: business-type detection, 6-dimension weighted scoring, and severity-ranked issue lists in one deliverable report structure.
- Parallel subagent design speeds up analysis with clearly separated domain responsibilities.
- Strict quality gates: 50-page cap, 30-second timeout, robots.txt compliance, rate limiting, and URL deduplication minimize load on the target site.
- Business-type-specific adjustments for SaaS, local, e-commerce, publisher, and agency sites, including per-type schema checklists.
- MIT-licensed and free, installed alongside an actively maintained 16-skill collection.
- Deeply relies on Claude Code's parallel subagent mechanism; porting to other platforms requires rework.
- All scores and insights come from LLM analysis with no automated test suite or benchmark validation, so scoring consistency is unverified.
- The 30-115% visibility lift cited in SKILL.md comes from a single 2024 study — a claimed figure, not measured by this tool.
- Each audit covers at most 50 pages, so large sites only get sampled conclusions.
- PDF reports, deep brand scanning, and monthly comparisons live in sibling skills; this skill itself does not produce PDFs.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| GEO Full-Site Audit this page | 48 · Use with care | ★ 11k | 20d ago | MIT |
| GEO-SEO AI Search Optimization Audit Skill | 43 · Not recommended | ★ 11k | 20d ago | MIT |
| llms.txt Analysis & Generation Skill | 46 · Use with care | ★ 11k | 20d ago | MIT |
| SEO Content Audit Skill | 48 · Use with care | ★ 2.8k | 7mo ago | MIT |
| GEO Crawlers Access Analysis Skill | 51 · Use with care | ★ 11k | 20d ago | MIT |
How did FollowSkills review this skill?
Declared tools include Bash, Write, and WebFetch — a fairly broad grant though not a red line; the workflow has explicit quality gates (50-page cap, 30s timeout, robots.txt respect, 1s rate limit) constraining and disclosing external crawling. Deducted for: no user confirmation before fetching targets, undefined Bash/Write usage, no rollback guidance, unverified publisher identity.
Orchestration is well structured with clear phases and prescribed error handling (log failures, continue, report in appendix); the repo includes SSR-detection tests for fetch_page fixing Issue #19. Deducted for: no test coverage of this skill's key paths (five subagent scorers, composite scoring), nothing executed in static review, and unverified consistency of scoring heuristics (e.g., 134-167 word optimal passages); capped at 10 statically.
Scenarios are clear (five business types with tailored adjustments), inputs/outputs well specified, trigger semantics explicit (/geo audit <url>). Deducted for: no Chinese-language support declared, brand scanning depends on platforms unreachable from mainland China (YouTube/Reddit/Wikipedia), and the core workflow depends on WebFetch reaching overseas sites — environment fit is limited.
Docs are well layered (purpose→workflow→scoring→output format→quality gates→business adjustments), MIT license explicit, repository structure clear. Deducted for: no version/changelog on the SKILL.md itself, subagent definition files not in evidence (hidden dependency), README mixed with heavy marketing and Skool community funnels, maintenance responsibility only informal.
Output format (GEO-AUDIT-REPORT.md) is thoroughly specified and directly usable; weighting and severity taxonomy are complete; marginal value (automated GEO audit) is plausible. Deducted for: category scores are subjective LLM judgments with no verified representative outputs; static review cannot confirm reports are client-deliverable as claimed; capped at 7 statically.
Auditable primary material exists (full SKILL.md, test file, Issue #19 reference) and the scoring formula is transparent. Deducted for: tests cover only fetch_page, no reproducible evidence for the five subagent scoring paths or composite GEO Score, README market statistics (+527%, 4.4x, 3x) lack citations, and the referenced Georgia Tech/Princeton study cannot be verified from the files.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →