Stop guessing what's holding your site back. Profile your site once. Then your AI agent runs the operation and you deploy what's ready:

  • Automate the audit: the full protocol across rendering, structured data, AI-citability, topical coverage and off-site, evidenced and severity-rated.
  • Automate the fix work: approved findings become a staged plan with ready-to-ship schema blocks, answer-first rewrites and config lines.
  • Automate measurement: Search Console reads, the GA4 AI-referral channel, and monthly checks on who ChatGPT, Perplexity & Google AI actually cite.
  • Spend nothing on tools: the whole loop runs on Search Console, GA4, PageSpeed and the free validators.

One download with your free customer account, free for commercial use. Open it with your agent and start the audit today.

How to set it up

  1. Download and unzip. You get a folder called ai-seo-starter: 18 plain-text files, no code, no installs, no paid tools.
  2. Open it with your AI agent. The workspace is agent-agnostic: Claude Code, Codex, Cursor, or whatever agentic system you already use; any agent that can read files can run it. In Claude Code the included CLAUDE.md orients the agent automatically; elsewhere, provide the files and say "read CLAUDE.md first".
  3. Fill in your site profile. Open context/site-profile.md and complete it, or tell the agent: "Interview me against the site profile template and fill it in." Existing material (your about page, a positioning doc, old keyword research) can go straight into the folder for the agent to distill; it confirms the result back to you.
  4. Run the audit. Workflow 01 is the full protocol: the rendering floor, structured data and entity, per-page AI-citability, topical coverage, off-site. The agent works phase by phase and tells you exactly what to paste where it can't fetch (View Source, PageSpeed numbers, Search Console reads). The report lands evidenced and severity-rated.
  5. Approve priorities and the fix plan writes itself. The agent turns your approval into a staged plan: schema blocks, answer-first rewrites, and config lines drafted ready-to-ship, with clear whose-move markers. You deploy; the plan tracks verification.
  6. Measure. The 28-day Search Console reads, the GA4 AI-referral channel, and monthly citation spot-checks: who do ChatGPT, Perplexity, and Google AI Mode actually cite for your queries? Logged, so every next cycle stands on evidence.

What this is

Search split in two. People still type queries into Google, and increasingly they ask an answer engine and read one synthesized answer that cites a few sources and never shows the rest. Winning now means two jobs at once: ranking and being the cited answer. This resource hands you a working agent workspace that runs both as one operation:

  • One site profile every specialist reads: your site, money pages, must-win queries, competitors, and baselines, so the audit has your priorities instead of generic best practice.
  • Three workflows: the website audit protocol (rendering floor → schema + entity → per-page citability → coverage → off-site), the fix plan (approved findings become a staged, tickable plan with every artifact drafted), and measure + log (the loop that tells you what actually moved).
  • Four specialists: the site auditor (evidence-disciplined findings), the page optimizer (the 10-point AI-citability check + answer-first rewrites that keep your voice), the schema engineer (facts-only JSON-LD with validation steps), and the topic researcher (the questions your buyers ask that no page of yours answers).
  • Fix plans as living state: checkboxes, whose-move markers, a verification ledger; any session picks up exactly where you stopped.

Why agent workflows, not a checklist

Because an audit is judgment applied to evidence, not a list to skim. These agent workflows give your agent a memory (your site profile, your findings log, your open fix plans) and a contract: every finding carries evidence, every artifact ships ready with its validation step, credibility gaps get flagged for you to fill with real facts (never padded with invented ones), and you deploy. That is the same architecture always-on SEO systems are built on: complete, yours, and ready to run this week.

What makes it different

  • The agent orients before acting. Every session starts with a workspace status: profile state, last audit, open fix plan and whose move it is.
  • Evidence over opinion. Findings name the URL, what was observed, and how to reproduce it; unevidenced suspicions are labeled hypotheses, never findings.
  • Both jobs, explicitly. Classic ranking signals and AI-citability are audited side by side, including your robots.txt posture toward AI crawlers, your entity's resolvability, and who the answer engines cite today for your queries.
  • Free tools only. Search Console, GA4, PageSpeed Insights, the schema validators, View Source, and the answer engines themselves. The workspace assumes nothing paid.

It's free

Free under CC-BY-4.0: use it, adapt it, and ship it commercially. Attribution appreciated.