Exécutez n'importe quel Skill dans Manus
en un clic

Exécutez n'importe quel Skill dans Manus en un clic

pre-launch-seo-check

Étoiles92

Forks9

Mis à jour19 juin 2026 à 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.

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Exécuter dans Manus

Source

spronta

spronta/crawlie

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Exécuter dans Manus

Métiers associésSOC

Basé sur la classification professionnelle SOC

Analystes en études de marché et spécialistes en marketingProfessions des affaires et des opérations financières·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)."

Plus depuis ce dépôt

même dépôt

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)."