Jonathon's AI Wiki

SEO & Content Ecosystem Architecture

6 min read

SEO & Content Ecosystem Architecture

Source: wiki synthesis: gsc-autonomous-seo, seo-audit-skill, clawdbot-competitive-intel, blog-agent-worker

Four projects at WEO Marketly form a connected SEO and content ecosystem. Each project handles one phase of the content lifecycle — opportunity detection, content generation, quality validation, and competitive monitoring — and feeds its outputs into the next. The long-term vision is a fully autonomous content optimization loop: detect opportunity, generate content, validate quality, publish, monitor impact, repeat.

Data Flow

GSC Autonomous SEO (per-query optimization)
     |
     | identifies underperforming queries/pages
     v
Blog-Agent-Worker / Pulse (generates/improves content)
     |
     | new or refreshed content
     v
SEOmator Audit (quality validation, 251 rules)
     |
     | validated content + audit score
     v
WordPress (publishing)
     |
     | published content generates search data
     v
GSC Autonomous SEO (monitors impact, restarts cycle)

     Clawdbot (competitive intelligence)
     |
     | competitive gaps, trending topics, market shifts
     +-----> feeds into Blog-Agent-Worker for content gap analysis
     +-----> feeds into GSC engine for query prioritization

The Four Components

1. GSC Autonomous SEO — Opportunity Detection

  • Reads GSC data at the per-query level
  • Scores opportunities using 5-component formula
  • Identifies WHICH queries on WHICH pages need improvement
  • Outputs: prioritized list of (query, page, enhancement type) triples

2. Blog-Agent-Worker (Pulse) — Content Generation

  • Takes opportunity briefs and generates or improves content
  • 7-agent pipeline: Research, Write, SEO, Edit, Social, Email
  • 117-point quality validation before handoff
  • Outputs: publication-ready content with derivatives

3. SEOmator Audit — Quality Validation

  • 251 rules across 20 categories validate content before and after publishing
  • Catches technical issues (schema errors, accessibility, performance) that content generation misses
  • Produces audit scores that feed back into opportunity scoring
  • Outputs: pass/fail report with category breakdowns

4. Clawdbot — Competitive Intelligence

  • Tracks 16 competitors across 8 channels monthly
  • Identifies content gaps (topics competitors cover that WEO does not)
  • Monitors competitive position changes to detect market shifts
  • Outputs: monthly reports, competitive gap analysis, trend alerts

Shared Infrastructure

Claude API

  • All four projects use Claude models (Opus, Sonnet, Haiku)
  • Model allocation pattern consistent across projects: Opus for creative/orchestration, Sonnet for analysis, Haiku for volume
  • Shared API key management via .env files

Railway Deployment

  • Blog-Agent-Worker deployed to Railway (production URL)
  • Clawdbot reports generated locally but could move to Railway
  • Hermes Agent on Railway as potential cron scheduler for all pipeline runs

Google APIs

  • GSC API v3 — Search performance data (GSC engine)
  • GA4 — Engagement metrics (Blog-Agent-Worker content health scoring)
  • Google Search Console — Crawl status monitoring (GSC engine cooldowns)

Data Stores

  • PostgreSQL — Per-query state, enhancement history (GSC engine)
  • SQLite — Content storage, generation history (Blog-Agent-Worker), audit scores (SEOmator)
  • Filesystem — Bot memory (Clawdbot), website baselines, place ID caches

Integration Opportunities

Currently Possible

  1. GSC engine identifies pages → Blog-Agent generates improvements → SEOmator validates → GSC monitors impact

    • The core loop. GSC engine output becomes Blog-Agent-Worker input. SEOmator validates before publish. GSC engine tracks post-publish performance
    • Cooldowns ensure the loop does not run faster than Google can evaluate changes

  2. Clawdbot competitive data → Blog-Agent content gap analysis → targeted content creation

    • Monthly competitive reports identify topics where competitors rank but WEO does not
    • Blog-Agent-Worker generates targeted content to fill those gaps
    • GSC engine monitors whether new content captures the identified queries

  3. SEOmator audit scores → GSC engine opportunity formula (audit-weighted scoring)

    • Pages with low audit scores AND high search potential get prioritized
    • This prevents the engine from optimizing content on technically broken pages

Future Integrations

  1. All feed into GoHighLevel via API (when scopes approved) for client-facing dashboards

    • Current GHL API token has limited scopes (locations, company, snapshots only)
    • When expanded: push competitive position data, audit scores, and content health into GHL dashboards
    • Dental practices see their SEO performance alongside CRM data

  2. Hermes Agent as cron scheduler for automated pipeline runs

    • Hermes (deployed on Railway) schedules weekly GSC data pulls, monthly competitive reports, and triggered content generation
    • Replaces manual node scripts/generate-monthly-report.js runs
    • Event-driven: GSC detects opportunity → triggers Blog-Agent → triggers SEOmator → publishes

  3. Cross-project content health dashboard

    • Unified view combining: GSC query performance, Blog-Agent quality scores, SEOmator audit scores, Clawdbot competitive position
    • Single pane of glass for “how is our content performing across all dimensions?”

The Vision

Fully autonomous content optimization loop:

  1. Detect — GSC engine identifies underperforming queries
  2. Analyze — Clawdbot provides competitive context for those queries
  3. Generate — Blog-Agent-Worker creates or improves content
  4. Validate — SEOmator runs 251-rule audit on the generated content
  5. Publish — WordPress REST API publishes validated content
  6. Monitor — GSC engine tracks impact with crawl-aware cooldowns
  7. Repeat — Loop restarts when cooldowns expire and new data is available

Human review remains at step 3 (content approval) and step 4 (audit review). The system is autonomous in detection and monitoring but human-gated in creation and validation. Hunter’s design principle: the human review toggle is never removed.

Key Takeaways

  • The four projects form a natural pipeline: detect, generate, validate, monitor
  • Shared infrastructure (Claude API, Railway, Google APIs) reduces operational overhead
  • The ecosystem’s value is greater than the sum of its parts — competitive data improves content targeting, audit scores improve prioritization, per-query tracking closes the feedback loop
  • Human review remains at content approval and audit review — full autonomy is the technical capability, not the operational policy
  • GoHighLevel integration is the bridge between internal tooling and client-facing visibility
  • Hermes Agent is the missing scheduler that turns manual pipeline runs into automated workflows

Try It

  1. Map the current state: which integrations between projects are manual vs automated?
  2. Start with integration #1 (the core loop) — it delivers the most value with the least new infrastructure
  3. Set up Hermes Agent cron scheduling for the GSC data pull (weekly) and Clawdbot report (monthly) as the first automation step
  4. Track integration progress in a shared dashboard — even a simple spreadsheet showing “which data flows are connected” reveals gaps
  5. When GHL API scopes expand, build the client-facing dashboard as the external interface to the internal ecosystem

Graph View

Backlinks