Die README konnte nicht geladen werden. Die Skills des Repositorys lassen sich unten weiterhin ansehen.
Six Agent Skills for reviewing marketing sites: one full review and focused checks for crawl access, factual claims, content, speed, and launches. Findings include evidence, severity, a proposed fix, and a way to verify it.
The author’s walkthrough describes a review of a local test site with deliberate errors: an article tagged noindex and a price mismatch between visible copy and structured data. The examples show the report’s evidence, explanation, fix, and verification format on a controlled test site.
Use blueline as the main entry for a site, page, draft, or pull request. A full review covers all ten laws; a code-change review checks what the diff can break. For a narrower task, choose blueline-crawl, blueline-claims, blueline-content, blueline-speed, or blueline-launch.
The README documents use with Claude Code and Codex. The skills need a shell and network access for live-site checks. Helper scripts need Node.js 22 or later; the references provide curl alternatives without Node. Supply the site, page, draft, or code change you want reviewed.
After installation, call /blueline <url> in Claude Code or $blueline <url> in Codex, replacing <url> with a site you own or have permission to review. The documented output gives a verdict for each law, then findings with evidence, a fix, and verification steps.
Script output contains facts and possible problems for the agent to check; it does not complete every review check. Requests are read-only and sequential, with a default crawl limit of 100 pages. The scripts do not apply robots.txt. Browser comparisons need a browser; real-user speed checks need CrUX access or an available PageSpeed Insights result. Missing evidence is marked “not verified”. The method does not guarantee rankings, AI citations, or traffic.
Loris Comba explains blueline, a collection of six marketing-site review Skills. Its checks connect findings to evidence, severity, proposed fixes and verification steps; the Article demonstrates noindex and price-mismatch findings on a deliberately faulty local test site.
Es werden 6 von 6 gesammelten Skills angezeigt.
Check that the facts and claims on a site or in a draft hold up, with evidence for every finding. Covers prices, requirements, FAQ answers, authors and dates stored once and matching the structured data, public numbers with their date and source, real quotes…
Quellsprache: Englisch
Review a page or a draft article so it wins the search it targets, with evidence for every finding. Covers one search intent per page, titles and H1s that lead with the query, descriptions, headings, answer-first writing, plain copy, pages that compete for…
Quellsprache: Englisch
Check whether search engines and AI assistants can read a site, with evidence for every finding. Covers what is in the raw HTML before JavaScript runs, head tags and structured data, real links, robots.txt and each AI crawler's access, one URL per page…
Quellsprache: Englisch
Check that a site can go live, or move, without losing its traffic, with evidence for every finding. Covers production indexable and staging not, robots.txt, the sitemap, canonicals, redirects for every old URL, real 404s, build checks that refuse broken…
Quellsprache: Englisch
Review a marketing website, a single page, a draft article or a code change against the ten laws of a marketing site, and prove every finding with evidence. Covers what search engines and AI assistants can read in the HTML, one URL per page, redirects and…
Quellsprache: Englisch
Find why a marketing page is slow, and what to fix first, with evidence for every finding. Covers Core Web Vitals from real-user data, render-blocking and third-party scripts, the JavaScript the page loads, dependencies, fonts, the main image, layout shifts,…
Quellsprache: Englisch
Diesen Prompt kopieren und in den verwendeten KI-Assistenten einfügen.
Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/loriscomba97/blueline.