Navin

SEO module - Overview

The SEO module (sidebar → SEO, route #/seo) is a senior SEO desk: grounded audits (live fetch), intent-mapped keywords, publish-ready content, schema, competitor gaps, links/local/GEO - under the expert contract. Without a data API (DataForSEO/Semrush), live Search Console (MCP), or a GSC export, volumes/KD stay qualitative (n/a - requires API data) and are never invented.

Connect Search Console

For live queries, pages, and index signals, connect the Google Search Console MCP preset:

  1. Open Settings → MCP.
  2. Install the Google Search Console (search-console) preset.
  3. Provide either an OAuth Desktop client_secrets.json path (GSC_OAUTH_CLIENT_SECRETS_FILE) or a service-account JSON path (GSC_CREDENTIALS_PATH, with GSC_SKIP_OAUTH=true).
  4. Requires uvx (Astral uv) on the machine that runs the gateway. Destructive sitemap tools stay off by default.

SEO skills prefer this MCP over CSV exports when it is connected. Cron/loop jobs reuse the same gateway env.

Evidence engine

/seo uses the built-in seo tool before interpretation. Its actions are crawl, audit, schema, psi, crux, serp_snapshot, serp_history, score, report, and pipeline. Crawls are bounded and reuse the secure Scraping transport. PSI v5 and CrUX v1 return an explicit data_gap without optional API credentials. SERP snapshots require DataForSEO or Semrush, include source and confidence, and are stored atomically as append-only history. No position, volume, or performance metric is inferred.

How it works

  1. Open SEO in the sidebar.
  2. Click an action card in one of the three groups - Audit, Research, Optimize (see Actions).
  3. The chat opens with /seo and the card prompt already in the composer. Add the site or topic in chat, then send.
  4. The agent delivers actionable output ordered by impact, saved as files when substantial.

Direct usage in an SEO module chat:

/seo https://example.com - full technical audit
/seo keyword research for artisan bakery in Lyon

The /seo command

Command/seo [url|topic]
LifecycleAgent workflow (runs a full agent turn)
Skills preloadedstudio-expert-contract, critic-reviewer, seo-technical-auditor, keyword-research, on-page-seo-optimizer, seo-content-writer, backlink-strategy, competitor-seo-analysis, local-seo, geo-ai-search-optimizer, seo-data-provider
BoardTracked run (live plan via project-board)
OutputFiles under seo/ + seo-report-*.html + expert gate

Coverage

AreaWhat the agent does
TechnicalBounded crawl evidence for status, redirects, robots, sitemaps, canonicals, hreflang, meta, headings, structured data, internal links, PSI and CrUX
KeywordsSeed expansion, intent classification, qualitative prioritization, plus provider metrics only when their source is available
ContentBriefs and full articles: title, H-structure, entities, internal links, FAQ, schema.org markup
CompetitorsContent strategy, site structure, targeted keywords, gap analysis with opportunities
LinksInternal mesh plans and backlink acquisition with outreach templates
LocalBusiness profile, citations, reviews, localized pages
AI search (GEO)Optimizing for AI-generated answers and answer engines

Continuous monitoring

Combine with Navin's autonomy features:

  • /goal monitor example.com rankings weekly and alert me on drops - a sustained goal the agent keeps pursuing.
  • Cron jobs (via the cron skill) for scheduled audits and reports.
  • The seo-monitoring and website-monitor skills for recurring checks.

Tips

  • Always give the URL when auditing; the agent fetches real pages.
  • Chain: technical audit → fix list → /forge (in Dev) to apply the fixes to your codebase.
  • Ask for the deliverable format you need: table, CSV file, or full article in markdown/HTML.