SEO

/seo — SEO Analysis & Optimization

Herramienta de análisis SEO completo para TC Insurance. Funciona despachando agentes especializados según el sub-comando.

Sub-comandos

Parse the first argument after /seo to determine which analysis to run:

Execution Flow

1. Parse Sub-command

Extract the sub-command from the skill args. If no args provided, show the help table above and stop.

2. Crawl Data Management (for audit, onpage, full)

Before dispatching agents that analyze the live site:

Check if docs/seo/data/crawl-{YYYY-MM-DD}.json exists (today's date)

If it does NOT exist:

1. Check if a dev server is running on localhost:3000 (lsof -i :3000)
2. If running: node scripts/seo-crawl.mjs --localhost
3. If NOT running: Ask the user whether to start dev server or use production URL
4. Wait for crawl to complete

If it exists: skip crawl, use existing data.

Store the crawl data path for agent prompts: CRAWL_PATH=docs/seo/data/crawl-{date}.json

3. Dispatch Agents

Use the Agent tool to dispatch the appropriate agent(s). Each agent is defined in .claude/agents/.

For /seo audit:

Dispatch seo-technical agent with prompt:
"Run in AUDIT mode. Crawl data is at {CRAWL_PATH}.
Perform a full Technical SEO audit following your audit mode instructions.
Save your report to docs/seo/reports/{YYYY-MM-DD}-audit.md"

For /seo onpage:

Dispatch seo-technical agent with prompt:
"Run in ON-PAGE mode. Crawl data is at {CRAWL_PATH}.
Perform a full On-Page SEO audit following your on-page mode instructions.
Save your report to docs/seo/reports/{YYYY-MM-DD}-onpage.md"

For /seo analytics:

Dispatch seo-technical agent with prompt:
"Run in ANALYTICS mode. No crawl data needed.
Audit the analytics setup following your analytics mode instructions.
Save your report to docs/seo/reports/{YYYY-MM-DD}-analytics.md"

For /seo content:

Dispatch seo-content agent with prompt:
"Perform a full Content SEO analysis following your instructions.
Save your report to docs/seo/reports/{YYYY-MM-DD}-content.md"

For /seo local:

Dispatch seo-local agent with prompt:
"Perform a full Local SEO audit following your instructions.
Save your report to docs/seo/reports/{YYYY-MM-DD}-local.md"

For /seo ai:

Dispatch seo-ai agent with prompt:
"Perform a full AI/GEO SEO analysis following your instructions.
Save your report to docs/seo/reports/{YYYY-MM-DD}-ai.md"

For /seo full:

1. Run crawl if needed (step 2)
2. Dispatch ALL agents IN PARALLEL using multiple Agent tool calls in a single message:
   • seo-technical (audit mode) with crawl path
   • seo-technical (onpage mode) with crawl path — Note: use a separate agent instance
   • seo-technical (analytics mode)
   • seo-content
   • seo-local
   • seo-ai
3. Once all agents complete, consolidate results into docs/seo/reports/{YYYY-MM-DD}-full.md:
   • Combined summary with overall score
   • Top P0 issues across all domains
   • Top P1 issues
   • Quick wins from each domain
   • Links to individual reports

For /seo report:

1. List all files in docs/seo/reports/ sorted by date
2. Read the most recent report(s)
3. If a previous report of the same type exists, compute a delta:
   • Issues fixed since last report
   • New issues found
   • Score change (improved/declined)
4. Present summary to user

4. Report Saving

Every agent saves its own report to docs/seo/reports/YYYY-MM-DD-{type}.md.

For /seo full, additionally create a consolidated report that links to all individual reports.

5. Present Results

After the agent completes:

1. Read the generated report
2. Present a concise summary to the user:
   • Score
   • Number of P0/P1/P2 issues
   • Top 3 quick wins
   • Link to full report file

Agent Reference

Key Files

• Crawler: scripts/seo-crawl.mjs
• Crawl data: docs/seo/data/crawl-*.json
• Reports: docs/seo/reports/*.md
• Site config: src/config/site.ts
• Project guide: CLAUDE.md

Notes

• Agents generate reports with suggested fixes but do NOT auto-apply changes
• The user decides which fixes to implement after reviewing the report
• Reports are saved for historical tracking — running the same audit again allows comparison
• The crawler uses Playwright (already a project dependency)
• WebSearch is used by content, local, and AI agents for keyword research and competitive analysis