What does this skill do, and when should you use it?
This is one sub-skill inside the zubair-trabzada/geo-seo-claude monorepo (16 bundled skills), located at skills/geo-citability/SKILL.md. Grounded in 2024 GEO research from Princeton, Georgia Tech, and IIT Delhi, it estimates how likely AI systems are to extract and quote passages from a page. It segments content into blocks and scores five weighted dimensions: Answer Block Quality (30%), Passage Self-Containment (25%), Structural Readability (20%), Statistical Density (15%), and Uniqueness (10%). The output is a GEO-CITABILITY-SCORE.md report with per-section scores and specific rewrites. Unlike keyword-driven SEO, it optimizes for extractability — how easily an AI can lift a passage as a direct answer.
Fetches the target URL via WebFetch and parses the main content (excluding nav, footer, sidebar, ads); splits the page into blocks at each H2/H3 heading; scores each block 0-100 on the five dimensions and computes a weighted page score; identifies the three strongest and three weakest blocks and calculates the share of blocks above 70 (citability coverage); generates specific rewrites for blocks scoring below 60 (answer-first openings, suggested statistics, structural fixes); writes a GEO-CITABILITY-SCORE.md report with a weighted score table, per-section breakdown, and quick-win recommendations with expected point lifts.
- Content marketing teams checking whether long-form articles contain passages that ChatGPT or Perplexity can quote verbatim before publishing
- GEO/SEO consultants delivering AI-visibility diagnostics to clients, using the score and rewrite list as the deliverable
- SaaS companies optimizing documentation so AI assistants cite their product when answering related questions
- Bloggers whose answers get buried mid-paragraph, restructuring toward answer-first openings per the suggestions
- E-commerce teams optimizing product pages to become citation sources for AI shopping recommendations
How do you install this skill?
- Static review only; no code was executed — all scores are conservative, low-confidence estimates.
- Citability scoring is fully LLM-subjective; repeated runs on the same page may yield different scores.
- Claimed research figures (134-167 word optimum, +30-115% visibility) lack checkable primary sources; verify before relying on them.
- The skill fetches arbitrary external URLs via WebFetch — mind target-site privacy/compliance; the output may overwrite an existing GEO-CITABILITY-SCORE.md.
- Repository tests do not cover this skill's path; quantified '+X point lift' promises should not be trusted.
- Shell / CLI
- Network access
- Local filesystem
Claude Code CLIPython 3.8+Git
Install the whole collection first: on macOS/Linux run curl -fsSL https://raw.githubusercontent.com/zubair-trabzada/geo-seo-claude/main/install.sh | bash, or git clone the repo and run ./install.sh; on Windows use install-win.sh from Git Bash. Requirements: Claude Code CLI, Python 3.8+, Git; Python dependencies go into an isolated venv at ~/.claude/skills/geo/.venv/. Note: the skill itself lives in skills/geo-citability/ — whether the installer enables this sub-skill individually is not documented in the source.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/zubair-trabzada/geo-seo-claude.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/geo-citability" ~/.claude/skills/
rm -rf "$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?
Within the installed collection, the README exposes /geo citability <url> in Claude Code. If invoking the skill directly, ask Claude Code to analyze a URL for AI citability; the skill will fetch the page via WebFetch and follow the five-dimension procedure in SKILL.md, writing GEO-CITABILITY-SCORE.md at the end. Allowed tools: Read, Grep, Glob, Bash, WebFetch, Write.
What are this skill's strengths and limitations?
- Fully transparent rubric: weights and per-band criteria for all five dimensions are written into SKILL.md and auditable
- Actionable output: rewrite suggestions are sentence-level per weak block, with estimated point lifts
- Includes platform-specific citation-preference reference for ChatGPT, Perplexity, Claude, Gemini, and Copilot
- Ships reusable high/low-citability example passages and self-containment checklists
- The research claims (134-167 word optimal passages, uplift percentages) are third-party findings relayed in the doc; the repo provides no test suite or validation data
- Scoring is heuristic and AI-judged; the same page can receive different scores across runs
- No persistent storage among the allowed tools — results live only in the generated report file
- Scope is citability scoring alone; crawler analysis, brand scanning, and PDF reports belong to sibling skills in the monorepo
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 Citability Scoring Skill this page | 48 · Use with care | ★ 11k | 20d ago | MIT |
| Blog GEO Citation Audit | 55 · Use with care | ★ 2.4k | 11d ago | MIT |
| GEO Content Quality & E-E-A-T Assessment | 47 · Use with care | ★ 11k | 20d ago | MIT |
| GEO Citation Optimizer | 52 · Use with care | ★ 3k | 11d ago | Apache-2.0 |
| GEO-SEO AI Search Optimization Audit Skill | 43 · Not recommended | ★ 11k | 20d ago | MIT |
The repo positions the whole collection against traditional SEO: classic SEO optimizes keyword density and engagement metrics, while this skill optimizes extractability — whether an AI system can pull a passage out and present it as a direct answer.
How did FollowSkills review this skill?
Uses only Read/Grep/Glob/Bash/WebFetch/Write, no destructive defaults, output is a local Markdown report, so main risks are visible; but the SKILL.md declares no confirmation mechanism, no URL scoping limits, no rollback or data-flow disclosure, and Write behavior (overwrites) is unspecified — deducted accordingly.
The analysis procedure (fetch, segment, weighted scoring, output format) is internally self-consistent and well structured, but scoring rests entirely on subjective LLM judgment with vague 0-100 band criteria; no test coverage of this skill's key path (existing repo tests target fetch_page SSR heuristics only), and failure handling for timeouts, non-HTML, or paywalled pages is undefined — deducted.
Target scenario (content teams/agencies assessing AI-citation readiness) is clear and the 'citability' trigger is semantically precise with a fixed output format; but capability boundaries (JS-rendered pages, non-English content) are undeclared and there is no Chinese-language support statement — deducted.
SKILL.md is well layered (rubric → procedure → output template → reference data) with compliant frontmatter and good examples; however there is no versioning/changelog, no known-limitations disclosure, cited research figures lack checkable sources, and maintenance responsibility is only implicit at repo level — deducted.
The rubric is concrete and the output template is directly deliverable; a static read suggests the core task (citability score plus rewrite suggestions) is plausibly achievable, but score reproducibility/accuracy is unverified, the quantified 'expected citability lift +X points' claims are unsupported, and marginal value over a careful manual review is limited — deducted.
Cites Princeton/Georgia Tech/IIT Delhi GEO research with specific percentages, but all are unverifiable secondhand restatements; the repo has tests and CI signals, yet none cover this skill's scoring path, leaving no independently reproducible conclusions — deducted.
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 →