website-intelligence

name	website-intelligence
description	Use this skill when the user asks to audit, analyze, score, or inspect a website for SEO, content, performance heuristics, tech stack, AI/GEO visibility, backlinks/search signals, conversion clarity, or competitor site intelligence. Triggers on phrases like audit this website, analyze site, website score, SEO audit, competitor website, site performance, tech stack, content audit, AI visibility, GEO visibility, and website intelligence. Use it to deploy the website analysis team for full audits or choose quick web_fetch/web_fetch_batch checks for narrow questions.
emoji	🧩
version	1.1.0
triggers	audit website, analyze site, website score, SEO audit, competitor website, site performance, tech stack, content audit, AI visibility, GEO visibility, website intelligence, site analysis, conversion audit, homepage audit

Website Intelligence

Full playbook for deploying the one-shot 5-agent website analysis team and interpreting results.

---

Tool

deploy_analysis_team({ url })

• url: Full URL including https:// (e.g. https://example.com)
• Spawns 5 specialist agents in parallel — runs async, result delivered to main chat via notification event
• Report saved to workspace/reports/ after completion
• Team self-deletes after delivery

---

When to Use

Use deploy_analysis_team when the user asks to:

• Audit their website
• Check SEO or search rankings
• Analyze site performance or tech stack
• Understand their AI/GEO visibility
• Research a competitor's website
• Get a site score or full report

For a quick single-question web check (e.g. "does my site have HTTPS?"), use web_fetch directly instead. For several known pages or competitor/source URLs, use web_fetch_batch.

---

What the 5 Agents Do

Agent	Checks	Output file
SEO Scanner	Title, meta description, H1/H2, image alt text, canonical, robots meta, SERP presence, top keywords	findings-seo.md
Performance & Stack Detective	Framework detection (Next.js, WP, Shopify), JS bundle size, render-blocking, mobile viewport, lazy loading	findings-performance.md
GEO (AI Visibility) Agent	Brand presence in AI-generated search results, featured snippets, knowledge panels, AI citation score	findings-geo.md
Backlinks & SERP Intelligence	Domain authority signals, top referring domains, competitive ranking gaps	findings-backlinks.md
Content Audit Agent	Page copy quality, readability, value proposition clarity, CTA strength, content gaps	findings-content.md

---

Workflow

Standard site audit

1. deploy_analysis_team({ url: "https://example.com" })
2. Agents run in parallel (async — takes 2-5 minutes)
3. Result delivered as team_event notification to main chat
4. Report compiled at workspace/reports/[domain]-analysis.md
5. Read report and summarize top 3 priority fixes for the user

Competitive research

1. deploy_analysis_team({ url: "https://competitor.com" })
2. When report arrives, compare against user's own site
3. Identify gaps: keywords they rank for, tech advantages, content angles
4. Output a gap analysis with 5 actionable items

---

Reading the Report

After the report arrives, structure your response as:

1. Score summary — give each area a rating (SEO: X/10, Performance: X/10, GEO: High/Medium/Low/Invisible, Content: X/10)

2. Top 3 critical issues — the highest-impact problems to fix first

3. Quick wins — things fixable in <1 day that have outsized impact (missing meta description, no alt tags, etc.)

4. Strategic recommendations — 2-3 longer-term moves (tech stack upgrade, content strategy, link building)

Keep it direct. Numbers and specifics over adjectives. If data is missing from a section, note it as "scraping limitation" rather than guessing.

---

API Limitations

The team uses web_fetch, web_fetch_batch, and web_search — no Lighthouse API or PageSpeed Insights API. This means:

• Performance scores are heuristic, not Lighthouse numbers
• Backlink data is search-signal based, not Ahrefs/Moz level
• GEO visibility is based on web_search sampling, not exhaustive

When doing manual follow-up research after the team report:

• Use web_search({ query, fetch_top_k: 2-5 }) for compact source discovery plus reading.
• Use web_fetch_batch for selected competitor pages, review pages, docs, or SERP evidence URLs.
• Use browser tools only for live interaction, JS-rendered content, screenshots, or visual QA.

For production-grade audits, recommend the user also run Google PageSpeed Insights and Ahrefs manually and share results for a combined analysis.

---

Changelog

Date	Change
2026-03-19	Initial skill created. Tool wired in Block A2.