A Practical GEO Audit Pipeline with CLI and MCP: What to Check Before AI Search Cites You

Description and use

Audit GEO sites with GeoScore to ensure AI bots can access, understand, and trust pages, boosting AI search visibility.

Problem

Web pages often fail to be trusted or cited by AI systems because they lack proper crawler access, consistent machine-readable identity, and trust signals, leading to poor AI search performance.

Outcome

Implementing GeoScore audits ensures AI systems can crawl, understand, and trust web pages, improving the likelihood of AI citations and overall SEO performance.

Workflow

• 1. Update robots.txt to allow GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended and add sitemap.
• 2. Verify live responses for representative URLs to ensure policy is served.
• 3. Check machine-readable identity: title/H1 match, canonical URL present, JSON‑LD matches visible copy, correct schema types, sameAs links.
• 4. Audit surrounding pages: /about, /blog/, /contact, /privacy, /terms, /sitemap.xml for trust signals and correct HTTP status.
• 5. Capture live HTTP headers (HSTS, CSP, etc.).
• 6. Run GeoScore CLI: npx geoscore https://example.com (or with --json, --quiet, --fix).
• 7. For automation, integrate GeoScore into CI: use --json to feed dashboards, --quiet for scripts, --fix to generate starter files.
• 8. Expose audit via MCP: configure MCP server with geoscore-mcp command, use audit_url and generate_fix_files tools.
• 9. Iterate: audit live URL → identify missing signal → edit page/server → rebuild/deploy → re‑audit.
• 10. Maintain a repeatable scorecard with six checks: crawler access, machine-readable identity, structured data, trust routes, technical trust, measurement.

Evidence

GeoScore ships as an open‑source CLI and MCP server. A basic audit can be run with: npx geoscore https://example.com

Source

The fix mode generates starter files for llms.txt, robots.txt, and JSON‑LD, which should then be reviewed by a human before deployment.

Source