Any tool can hand you two hundred SEO problems. Knowing which six to fix first, what they're worth, and whether the fix worked. That's the whole job. Six nodes, thirty-two specialist checks, yours to download and run. Underneath, every site is the same six steps. Pick a business type below and watch what actually changes.
Three parts carry your business. Three carry the method, and those never change.
Run the full six-node SEO loop on my site and give me one self-contained HTML dashboard at the end. Do not stop between nodes to ask me questions. Make the reasonable call, state it in one line, and keep going. CONTEXT Business: [name] · [what you sell] Site: [url] · Market: [country] · Currency: [INR] What a win is worth: [average order value or qualified lead value] The decision I am trying to make: Which six things do I fix first, and what are they worth? Period: last 90 days SKILLS · LOAD THESE FIRST Before you run a single node, load these from my account. They carry my judgment; without them you are guessing: · seo-growth-os: the staged pipeline and its approval gates · seo-audit: the crawl, the fan-out, the scoring weights If any one of them is NOT available in this session, STOP and tell me exactly which skill is missing, before you run the node that needs it. Do not silently fall back to generic best practice. Name the gap, say what you will use instead, and let me decide whether to continue. NODE 1 · CONTEXT (who this is) Detect the business type from homepage signals: SaaS, local service, e-commerce, publisher or agency. Note the market, the buyer, and what a conversion is worth. This decides which specialists switch on later, so settle it first. → Write file: 01-Discover/Context.md NODE 2 · OUTBOUND (data from outside) Pull: competitor sites · the SERPs for my money terms · what buyers actually type · the category's content patterns. Ask me to confirm the competitor set before you analyse anything. State only what is true. No recommendations in this node. → Write file: 01-Discover/Research-Brief.md NODE 3 · INBOUND (data from inside) Crawl up to 500 pages once, respecting robots.txt, and cache every page. Every specialist reads from that cache. None of them re-fetch the site. Run the seven core checks always. Add conditional specialists only where the business type from Node 1 or a connected data source justifies them. Score 0–100. If a category could not be assessed, say so and redistribute its weight. Never score a category on guessed data. → Write file: 02-Audit/Audit-Report.md NODE 4 · DIAGNOSE (my rules run here) This is the node that does the work. Two passes, in order: 4a · Root causes. Sort every finding into Broken (exists but wrong), Missing (should exist, doesn't), or Leaking (works, underperforms). Collapse findings that share a cause into ONE gap. Thirty pages with the same invalid markup is one gap, not thirty. Count causes, not symptoms. → Write file: 03-Diagnose/Gap-Inventory.md 4b · Sequence. Score every gap: Impact × Effort × Confidence × Time-to-Impact. Sort into Quick Wins, Big Bets, and Kill/Defer. Order the dependencies: a markup fix that unblocks rich results comes before the content that needs them. Every item traces back to the finding that produced it. → Write file: 04-Prioritize/Roadmap.md Wait for my yes before anything gets built. NODE 5 · EXECUTE (drafts only) Work the roadmap top-down. Produce content briefs, ready-to-paste structured data, comparison pages, scale-page templates. Everything is a draft. Flag what a human must verify: facts, brand claims, internal URLs. Nothing publishes by surprise. → Write files under: 05-Execute/ NODE 6 · SCHEDULE (it runs without me) Capture a baseline the day fixes ship. Then set up a recurring weekly task that re-runs Nodes 3 and 4 from a cold start and delivers the result. The scheduled prompt must carry its own full instructions. A scheduled run remembers nothing about today. Answer two questions separately each week: did anything break, and are we trending up. Attach a leading indicator to every executed item: something I can check myself without commissioning another audit. Notify me only when it is done, or when something is wrong. → Write files: 06-Monitor/Baseline.md, 06-Monitor/Drift-Log.md, 07-Report/ FINAL OUTPUT · ONE HTML A single self-contained .html file. Dark. Brand colours #0E0E0E obsidian and #FF5C1A signal orange. It must show the PROCESS, not just the answer: · the six-node flow at the top, with what each node produced · per node: the sources used, the finding, and the evidence behind it · the gap inventory and the ranked roadmap · an approval checklist with real checkboxes · the schedule receipt: task name, cadence, next run · a closing "what this could NOT check" section No CDN, no external images, no server. It must open on a laptop with no internet. RULES · Never invent a number. · If a data source isn't connected, skip the check and mark it incomplete. · Match the same date range across every source, or the timing gap reads as a false trend. · Nothing goes live without my explicit yes. · Lead with the answer. No jargon without a number next to it. · If a skill above could not be loaded, say so at the top of the output, not buried at the bottom.
Nothing here is aspirational. Each one is wired into at least one stage of the pipeline. Twenty-two run the moment you download it. The rest need a data source connected, and four of those are free accounts you already own.
A local business needs geo-grid coverage. A store needs product markup. A SaaS needs comparison pages. Different business, different sources. Same six nodes.
Written down one time. Applied identically to every client, and to every engagement we take on later. It's why two audits six months apart are actually comparable.
The client context is cheap and changes weekly. The judgment is expensive and took years. Only one of them should be retyped for every new engagement.
Every stage writes a file, and they cross-reference each other, so any recommendation traces back to the evidence that produced it, and forward to the metric that proves it.
If a data source isn't connected, the check is skipped and marked incomplete, never estimated. A category we couldn't measure is stated plainly and its weight redistributed, so the score you see is the score we actually earned.
Each carries the observation it rests on, what it depends on, an explicit "here's how we'd know this failed", and a leading indicator you can check yourself. Advice you can't falsify isn't advice. It's decoration.
Everything writes to a linked record that persists. Re-engage in six months and the system knows what was found, what was fixed, and what regressed. No "let's start with a fresh audit" tax.
Four of the thirty-two components exist purely to answer whether AI Overviews, ChatGPT, Perplexity and Copilot can find, read and cite you, and whether that's improving month over month.
Two hard gates: you confirm the competitor set before analysis, and you approve the roadmap before anything is built. Every deliverable arrives as a draft with the human checks flagged.
Everything marked in terracotta changes with the site. Everything in mint is the method, and it took years to write down. Download it, run it on your own machine, keep whatever it finds. The audit was never the hard part.
Enter your email and the full bundle arrives in your inbox: 32 skills, a README, and a setup guide. It runs on your own machine.
The bundle is on its way to . It arrives as a zip attachment, so give it a moment. If it has not landed in a few minutes, check spam or tell us.