NikitasT2003

Serpentine — Data skill for Claude Code

Data community

An SEO skill for AI agents that writes the code instead of the PDF.

How to install Serpentine

This entry records only its repository, not the path inside it, so there is no exact command to give. Open NikitasT2003/seo-operator and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Serpentine does

An SEO skill for AI agents that writes the code instead of the PDF. Detects your framework, ships titles, canonicals, sitemaps, structured data and internal links into the repo, then curls the build to prove they arrived.

Alternatives in Data

HTML Anything — ✨ The agentic HTML editor — your local AI agent writes the HTML, you ship it 8.5k ★
LLM Redteam — LLM red-team corpus runner — fires categorized prompt-injection / jailbreak / system-prompt-leak / data-exfil 4.5k ★
Portaljs Architect — Recommend a data-portal architecture (storage, compute, catalog, access, hosting, metadata) from your needs, t 2.3k ★

README

serpentine

[![MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE) [![Validate](https://github.com/NikitasT2003/serpentine/actions/workflows/validate.yml/badge.svg)](https://github.com/NikitasT2003/serpentine/actions/workflows/validate.yml) [![Google guidance verified 2026-08-26](https://img.shields.io/badge/google%20guidance-2026--08--26-brightgreen)](references/crawl.md)

There is a SERP in the name. Everything else here is serious.

The standard SEO deliverable is a 40-page PDF that says "improve your meta descriptions."

This one improves your meta descriptions.

Install

git clone https://github.com/NikitasT2003/serpentine ~/.claude/skills/serpentine

Then say "do SEO on this site." Use `.claude/skills/` instead to scope it to one project.

What it does

1. Detect     framework, where  is controlled, SSR or client-only
2. Inventory  every route: title, description, canonical, H1, target
3. Target     one query per page, no two pages chasing the same one
4. Implement  crawlability, metadata, on-page, schema, links, CWV
5. Verify     curl the built output and confirm the tags are really there

Every step ends in a file changed or a stated reason no change was needed. It is not allowed to finish by suggesting you consider looking into things.

Step 5 exists because metadata has a habit of being extremely present in your components and completely absent from your HTML.

Things it finds in the first ten minutes

**The `noindex` from staging.** Added in March so the preview site would not get crawled. Shipped to production. Has been quietly instructing Google to ignore the entire website ever since. Nobody can see it in a browser, which is why it is still there.

**`X-Robots-Tag` in `vercel.json`.** Same idea, except it overrides your HTML and appears in no template, so grepping the components finds nothing and everyone concludes SEO is mysterious.

**The `robots.txt` paradox.** You added `noindex` to remove a page.