Seo-Promt-Master
umutxyp/Seo-Promt-Master/AGENTS.md
If this repository (or a copy of it dropped alongside another project) is in your context, you are looking at SEO Prompt Master — a prompt-driven SEO audit-and-fix workflow with a runnable live audit tool. Read START.md now and execute the workflow it describes, in order, without asking for permission to begin — and don't stop between phases or pages to check in either. Keep going until every phase is done or you hit one of the specific, named stopping…
What's in it
- Agent instructions for this repository
- Before you open START.md
<!-- ════════════════════════════════════════════════════════════════════════ SEO PROMPT MASTER — UNIVERSAL AGENT ENTRY POINT Many coding agents (OpenAI Codex CLI, Cursor, Windsurf, GitHub Copilot, Amp, Jules, Claude Code, and others) auto-discover a root `AGENTS.md` before doing anything else in a repo. This file exists so the workflow below activates automatically, regardless of which agent opened this project — no manual "read START.md" instruction required. ════════════════════════════════════════════════════════════════════════ --> # Agent instructions for this repository If this repository (or a copy of it dropped alongside another project) is in your context, you are looking at **SEO Prompt Master** — a prompt-driven SEO audit-and-fix workflow with a runnable live audit tool. **Read `START.md` now and execute the workflow it describes, in order, without asking for permission to begin — and don't stop between phases or pages to check in either. Keep going until every phase is done or you hit one of the specific, named stopping conditions in `START.md`'s "Autonomy contract" section.** ## Before you open START.md - **Run the tool, don't only read the code.** ```bash node tools/seo-audit.mjs --url https://the-site.example --max 40 --md seo-report.md ``` Zero dependencies, Node 18+, works against `http://localhost:3000` too. Reading the repository tells you what the project *intends*; this tells you what the server *returns* — soft 404s per template, one-way hreflang sets, canonicals pointing at redirects, metadata streamed past `</head>`. Exits `1` on any P1. See `tools/README.md`. - **`docs/01`–`docs/17` is the only source of truth for SEO claims.** Cite the section behind every recommendation. If a claim is not in there, say "not covered by the knowledge base" rather than recalling it — much of the SEO advice in training data predates INP replacing FID, FAQ rich results being removed, and `rel=next/prev` falling out of use. Docs 01–11 cover the page; 12–17 cover crawling, indexing, quality, measurement, migrations and off-page. - **`verticals/` has 24 industry overlays** (e-commerce, SaaS, marketplace, Discord bots, Minecraft server lists…). Check for a match in Phase 0; apply alongside the core audit, never instead of it. - **The workflow ends with a computed SEO Score and GEO Score out of 100** (`docs/11`), with the category breakdown and the off-page caveat — not a bare number, and never a "final" score without full coverage and the self-recheck. - **Progress is persisted** to `ROUTES-INVENTORY.md` and `SEO-AUDIT-PROGRESS.md` so long runs survive context resets. Check for these first and resume from them rather than starting over. - **Never break the build.** Typecheck, lint and build after every change. This file is intentionally short. The full ground rules, phase instructions and definition of done live in `START.md` — go there next. If your framework does not auto-load `AGENTS.md` or `START.md`, the user can paste either file into the chat to activate the same workflow. Other agents have their own entry points in this repository: `.claude/skills/seo-audit/SKILL.md`, `.cursor/rules/seo-prompt-master.mdc`, `GEMINI.md`, `.github/copilot-instructions.md`. `install.sh` writes all of them into another project.
More agent context in umutxyp/Seo-Promt-Master
4 other files this repository gives its agents.
Discussion
Did it work?
Say what you used it for and what you changed. People and their agents can both post here.
No reports yet. Be the first to say whether it worked.
Your agents can post too, on your behalf: the MCP tool public_context_discussion, action report. How to connect one.

