Geo Diagnose Agent Skill

Diagnose a brand, website, or page for evidence-backed GEO gaps and opportunities from user-supplied URLs, HTML, or evidence. Use for brand diagnosis, website or page audits, GEO gap analysis, and 品牌诊断、网站诊断、页面诊断. Excludes live AI-platform recall, ranking, and citation-share measurement.

5k tokens

context cost

the whole folder, loaded on every use

files

ships runnable scripts

copies elsewhere

how many repositories repackaged it

101

stars on the repo

on the repository, not the skill itself

Install

one command, takes just this skill from the repository

npx skills add https://github.com/yaojingang/GEOHub --skill geo-diagnose

What comes with it

16 439 bytes besides the instruction

agents/interface.yaml

evals/output/cases.jsonl

evals/semantic_config.json

evals/trigger_cases.json

manifest.json

references/audit-catalog.md

references/diagnosis-method.md

references/input-example.json

reports/output_quality_scorecard.md

reports/skill-ir.json

reports/trust-report.md

scripts/run_diagnose.py

The instruction itself

4 sections, as written by the author

GEO Diagnose

Workflow

Read references/diagnosis-method.md; read references/audit-catalog.md for non-legacy audit execution.
Prepare the diagnosis brief contract.
Run python3 scripts/run_diagnose.py --input <brief.json> --output <runs-root> --execution-mode <legacy|deterministic|research|provider>. Legacy remains the compatibility default.
Inspect quality-report.json, run-lineage.json, source_status, audit components, and limitations before using findings.
Deliver the complete Artifact Bus run directory.

Output contract

Return <runs-root>/<run-id>/ containing the normalized input and replayable HTML snapshots under input/sources/, structured diagnosis, deterministic Markdown report, evidence-linked remediation query map, opportunity map, quality report, run lineage, and run manifest. Non-legacy diagnosis adds versioned audit results, reconstructable score components, execution status, and a semantic digest to diagnosis.json.

Boundaries

Fetch only explicit public HTTP(S) canonical URLs without query strings. Accept remote HTML/XHTML only and keep unavailable or unsupported sources as source_gap. Every observed, provided, or inferred finding carries a content-derived evidence ID. Provider audit mode has no packaged adapter and records a degraded deterministic fallback. Never claim live AI-platform recall, ranking, or citation share.

How to use it

Copy the folder

Take yaojingang/geo-diagnose from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.