Ejecuta cualquier Skill en Manus
con un clic

Ejecuta cualquier Skill en Manus con un clic

pre-launch-seo-check

Estrellas92

Forks9

Actualizado19 de junio de 2026 a las 12:06

Run a pre-launch / pre-deploy SEO gate on a site or set of pages with crawlie, returning a clear pass/fail. Use when the user is about to launch, deploy, or ship and wants to verify there are no blocking SEO issues (broken links, 5xx, missing titles/canonicals, noindex slips), or wants a CI gate for SEO regressions. Works even if crawlie is not installed.

Instalación

Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.

Ejecutar en Manus

Fuente

spronta

spronta/crawlie

Abrir repositorio de GitHub Ver repositorios del creador

Descarga

Ejecutar en Manus

Ocupaciones relacionadasSOC

Basado en la clasificación ocupacional SOC

Analistas de investigación de mercados y especialistas en marketingOperaciones empresariales y financieras·SOC 13-1161

SKILL.md

readonly

name	pre-launch-seo-check
description	Run a pre-launch / pre-deploy SEO gate on a site or set of pages with crawlie, returning a clear pass/fail. Use when the user is about to launch, deploy, or ship and wants to verify there are no blocking SEO issues (broken links, 5xx, missing titles/canonicals, noindex slips), or wants a CI gate for SEO regressions. Works even if crawlie is not installed.

Pre-Launch SEO Check (powered by crawlie)

A go / no-go gate: crawl the target, apply blocking thresholds, and report a clear verdict with the exact blockers. Built for "are we safe to ship?" moments and CI. crawlie ships on npm as crawlie — no pre-install, no source repo needed.

Bootstrap — make crawlie runnable

Prefer mcp__crawlie__* tools if present this session.
Else CLI prefix: crawlie if on PATH, otherwise npx -y -p crawlie crawlie (auto-installs from npm; needs only Node).
If neither is available, ask for Node (https://nodejs.org) or the crawlie MCP, then stop.

<crawlie> below = your chosen prefix (or the matching MCP tool).

Workflow

Scope the gate. Either:
- the whole (staging) site: <crawlie> crawl <url> ..., or
- the critical pages only (home, pricing, signup, top landing pages): <crawlie> audit <url1> <url2> ... — faster, ideal for a tight gate.
Run with a failing threshold. For the crawl form, let crawlie set the exit code:
```
<crawlie> crawl <url> --format json --fail-on error
```
- --fail-on error → non-zero exit if any error-severity findings exist.
- --fail-on warning → stricter: non-zero on warnings too.
- In CI, that exit code alone gates the pipeline. (MCP: call crawl_site, then apply the threshold yourself from summary.errors / summary.warnings.)
Classify against launch blockers. Treat these as hard fails:
- broken-link, client-error, server-error, connection-error — dead links / pages.
- redirect-chain on primary URLs — equity + crawl-budget loss.
- title-missing, canonical-missing, mixed-content, not-secure, blocked-by-robots on pages that must be indexable — silent ranking killers. Treat thin-content / metadata-length / GEO findings as warnings (note, don't block), unless the user says otherwise.
Verdict. State PASS or FAIL up front.
- FAIL → list each blocker: page · rule · what's wrong · the fix (<crawlie> explain <rule-id> for guidance).
- PASS → confirm, and surface the Health + GEO scores plus any non-blocking warnings worth cleaning up post-launch.

CI snippet (offer if they want it automated)

npx -y crawlie crawl "$SITE_URL" --fail-on error --quiet

Non-zero exit fails the job. Pin a version (crawlie@<version>) for reproducible runs.

Make it permanent (optional — offer once, don't block)

"Want crawlie installed for good? npm i -g crawlie. For first-class agent tools, also claude mcp add crawlie crawlie-mcp (active after a Claude restart)."

Más de este repositorio

mismo repositorio

fix-broken-links

spronta/crawlie

Find and fix broken links, dead pages, and redirect chains on a website with crawlie. Use when the user asks to find broken links, 404s, dead links, link rot, redirect chains/loops, or 4xx/5xx pages, or to check that a site's links are healthy. Works even if crawlie is not installed.

2026-06-1992

geo-ai-readiness

spronta/crawlie

Assess how ready a website is to be cited and answered by AI search and LLMs (Generative Engine Optimization / GEO) using crawlie. Use when the user asks about AI search, generative engine optimization, GEO, getting cited by ChatGPT / Perplexity / Google AI Overviews, llms.txt, structured data for AI, or whether their content is "answerable" by LLMs. Works even if crawlie is not installed.

2026-06-1992

seo-site-audit

spronta/crawlie

Run a complete technical SEO + AI-search audit of a website with crawlie. Use when the user asks to audit, crawl, or check a site for SEO issues, broken links, redirects, missing titles/descriptions/metadata, canonical or heading problems, image alt text, duplicate or thin content, or generative-engine (GEO / AI-search) readiness. Works even if crawlie is not installed.

2026-06-1992

name	pre-launch-seo-check
description	Run a pre-launch / pre-deploy SEO gate on a site or set of pages with crawlie, returning a clear pass/fail. Use when the user is about to launch, deploy, or ship and wants to verify there are no blocking SEO issues (broken links, 5xx, missing titles/canonicals, noindex slips), or wants a CI gate for SEO regressions. Works even if crawlie is not installed.

Pre-Launch SEO Check (powered by crawlie)

Bootstrap — make crawlie runnable

Prefer mcp__crawlie__* tools if present this session.
Else CLI prefix: crawlie if on PATH, otherwise npx -y -p crawlie crawlie (auto-installs from npm; needs only Node).
If neither is available, ask for Node (https://nodejs.org) or the crawlie MCP, then stop.

<crawlie> below = your chosen prefix (or the matching MCP tool).

Workflow

Scope the gate. Either:
- the whole (staging) site: <crawlie> crawl <url> ..., or
- the critical pages only (home, pricing, signup, top landing pages): <crawlie> audit <url1> <url2> ... — faster, ideal for a tight gate.
Run with a failing threshold. For the crawl form, let crawlie set the exit code:
```
<crawlie> crawl <url> --format json --fail-on error
```
- --fail-on error → non-zero exit if any error-severity findings exist.
- --fail-on warning → stricter: non-zero on warnings too.
- In CI, that exit code alone gates the pipeline. (MCP: call crawl_site, then apply the threshold yourself from summary.errors / summary.warnings.)
Classify against launch blockers. Treat these as hard fails:
- broken-link, client-error, server-error, connection-error — dead links / pages.
- redirect-chain on primary URLs — equity + crawl-budget loss.
- title-missing, canonical-missing, mixed-content, not-secure, blocked-by-robots on pages that must be indexable — silent ranking killers. Treat thin-content / metadata-length / GEO findings as warnings (note, don't block), unless the user says otherwise.
Verdict. State PASS or FAIL up front.
- FAIL → list each blocker: page · rule · what's wrong · the fix (<crawlie> explain <rule-id> for guidance).
- PASS → confirm, and surface the Health + GEO scores plus any non-blocking warnings worth cleaning up post-launch.

CI snippet (offer if they want it automated)

npx -y crawlie crawl "$SITE_URL" --fail-on error --quiet

Non-zero exit fails the job. Pin a version (crawlie@<version>) for reproducible runs.

Make it permanent (optional — offer once, don't block)

"Want crawlie installed for good? npm i -g crawlie. For first-class agent tools, also claude mcp add crawlie crawlie-mcp (active after a Claude restart)."