Claude Skill

seo-technical-auditor

Audit technical SEO - indexation, sitemap, robots.txt, Core Web Vitals hints, redirects, canonical tags, and crawl errors. Use when a site underperforms in search or before a relaunch.

LLM Mart · 0 points · 0 0 0

Virus-scanned Reviewed automatically before listing.

Full trust report

Download navinspire-ia-navin-navin_skills_seo-technical-auditor-e9c73a3.zip · 3 KB

navinspire-ia/navin 31 4 forks AGPL-3.0 Updated 4d ago

Part of navinspire-ia/navin — 182 skills

Install

skills CLI npx skills add https://github.com/Navinspire-ia/navin/tree/main/navin/skills/seo-technical-auditor

Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install navinspire-ia-navin@llmmart

Git git clone https://github.com/Navinspire-ia/navin.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole navinspire-ia/navin collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

SEO Technical Auditor

Find and prioritize technical issues that block ranking. Evidence first, checklist second. Operate like a senior technical SEO: grounded fetches, impact × effort triage, no invented CWV scores.

When to use

Full site technical audit before relaunch or after traffic drops
Indexation / crawlability suspicion
Pre-migration checklist

When not to use

Pure content strategy with no URL (use keyword-research / seo-content-writer)
Paid ranking reports without API access (use MCP search-console if connected, else seo-data-provider or ask for GSC/PSI exports)

Audit checklist

Area	What to check	How
Indexation	noindex, canonical conflicts, `site:` coverage	fetch pages + web_search `site:domain`
robots.txt	blocked paths, sitemap line, crawl-delay	`web_fetch https://domain/robots.txt`
Sitemap	valid XML, fresh URLs, sample 404s	fetch `/sitemap.xml` (+ index children)
Redirects	chains, http→https, www, trailing slash	fetch variants; note status chain
Meta / head	title, description, robots, canonical, hreflang	sample 5-15 money pages
Structured data	JSON-LD presence/type	parse `<script type="application/ld+json">`
Internal links	orphans, deep money pages	sample nav + footer + in-content
Performance	LCP/CLS/INP hints only	ask for PSI/CrUX URL or note "requires PSI"
HTTPS	mixed content signals	fetch http upgrade + page assets notes
Hreflang	reciprocal FR/EN/AR	head tags on language variants

Workflow

Run seo action pipeline first for a full audit, or compose crawl, audit, schema, psi, crux, score, and report.
Treat engine findings and their evidence objects as the source of truth.
Use serp_snapshot only with DataForSEO or Semrush credentials and retain source/confidence.
Preserve every data_gap; do not replace missing PSI, CrUX, ranking, position, or volume data with estimates.
Add expert interpretation and prioritize findings as Critical / Important / Nice-to-have.
The canonical health score comes from seo action score. For legacy imported findings only, the helper remains available:

python navin/skills/seo-technical-auditor/scripts/audit_score.py findings.json

Or from workspace: python <skill_dir>/scripts/audit_score.py seo/audit-findings.json

Save report under seo/technical-audit-<domain>-<date>.md and feed seo-report-*.html.

Finding severity

Severity	Meaning	Examples
Critical	Blocks crawl/index or money page	sitewide noindex, sitemap 404, https broken
Important	Material ranking/UX drag	canonical loops, missing titles, thin money pages
Nice to have	Polish	missing FAQ schema, minor meta length

Report format

## Technical SEO Audit - <domain>
Date / scope / pages sampled

### Critical (blocks ranking)
1. issue - URL - evidence - fix - effort (S/M/L)

### Important
...

### Nice to have
...

### Quick wins (this week)
- ...

### Data gaps
- Core Web Vitals: requires PageSpeed Insights / CrUX
- Index coverage: prefer MCP `search-console` when connected; else GSC export if site: is inconclusive

Rules

Prefer live GSC via MCP search-console (Settings → MCP) for index coverage, queries, and page performance when available; otherwise ask for a CSV export or mark the gap.
Every finding needs a URL + observed evidence (status, snippet, header).
Never invent Lighthouse, CrUX, volume, or position data; use the engine's explicit data_gap.
Prioritize by impact × effort; do not dump 100 undifferentiated issues.
Sample-based audit is honest: state sample size, not "full crawl of N pages" unless you crawled them.
Pair with on-page-seo-optimizer for page-level rewrites and seo-monitoring for tracking.

Anti-patterns

Auditing from memory without fetching the live site
Claiming "Core Web Vitals fail" without data
Mixing content strategy opinions into technical Critical

Files (navin)

scripts

audit_score.py 2.7 KB

#!/usr/bin/env python3
# Copyright (c) 2026-present Navinspire IA
# SPDX-License-Identifier: AGPL-3.0-only

"""Score SEO audit findings by impact × effort for prioritization.

Input JSON: list of objects or {"findings": [...]} with keys:
  id/title, severity (critical|important|nice|critical|...),
  impact (1-5), effort (1-5), url (optional)

Prints a ranked markdown table and a 0-100 health score.
"""

from __future__ import annotations

import json
import sys
from pathlib import Path

SEVERITY_WEIGHT = {
    "critical": 5,
    "blocker": 5,
    "important": 3,
    "major": 3,
    "nice": 1,
    "nice_to_have": 1,
    "minor": 1,
}


def _load(path: Path) -> list[dict]:
    data = json.loads(path.read_text(encoding="utf-8"))
    if isinstance(data, dict):
        data = data.get("findings") or data.get("issues") or []
    if not isinstance(data, list):
        raise SystemExit("JSON must be a list of findings or {findings: [...]}")
    return [row for row in data if isinstance(row, dict)]


def _score(row: dict) -> float:
    sev = str(row.get("severity", "important")).lower().replace(" ", "_")
    impact = float(row.get("impact") or SEVERITY_WEIGHT.get(sev, 3))
    effort = float(row.get("effort") or 3)
    effort = max(effort, 0.5)
    return (impact * SEVERITY_WEIGHT.get(sev, 3)) / effort


def main() -> None:
    if len(sys.argv) < 2:
        print("Usage: audit_score.py findings.json", file=sys.stderr)
        raise SystemExit(2)
    path = Path(sys.argv[1])
    findings = _load(path)
    if not findings:
        print("No findings.")
        return

    ranked = sorted(findings, key=_score, reverse=True)
    critical = sum(
        1
        for f in findings
        if str(f.get("severity", "")).lower() in {"critical", "blocker"}
    )
    # Health: start 100, subtract weighted open issues (cap floor 0).
    penalty = 0.0
    for f in findings:
        sev = str(f.get("severity", "important")).lower().replace(" ", "_")
        penalty += SEVERITY_WEIGHT.get(sev, 3) * 4
    health = max(0, min(100, int(100 - penalty)))

    print("# Audit priority score\n")
    print(f"- Findings: {len(findings)}")
    print(f"- Critical/blocker: {critical}")
    print(f"- Health score (heuristic): {health}/100\n")
    print("| Priority | Title | Severity | Impact | Effort | URL |")
    print("|----------|-------|----------|--------|--------|-----|")
    for i, row in enumerate(ranked, 1):
        title = str(row.get("title") or row.get("id") or row.get("issue") or "finding")
        sev = str(row.get("severity", ""))
        impact = row.get("impact", "")
        effort = row.get("effort", "")
        url = str(row.get("url") or "")
        print(f"| {i} | {title} | {sev} | {impact} | {effort} | {url} |")


if __name__ == "__main__":
    main()

SKILL.md 4.3 KB

---
name: seo-technical-auditor
description: Audit technical SEO - indexation, sitemap, robots.txt, Core Web Vitals hints, redirects, canonical tags, and crawl errors. Use when a site underperforms in search or before a relaunch.
metadata: {"navin":{"emoji":"🔧","category":"seo"}}
---

# SEO Technical Auditor

Find and prioritize technical issues that block ranking. Evidence first, checklist second. Operate like a senior technical SEO: grounded fetches, impact × effort triage, no invented CWV scores.

## When to use

- Full site technical audit before relaunch or after traffic drops
- Indexation / crawlability suspicion
- Pre-migration checklist

## When not to use

- Pure content strategy with no URL (use `keyword-research` / `seo-content-writer`)
- Paid ranking reports without API access (use MCP `search-console` if connected, else `seo-data-provider` or ask for GSC/PSI exports)

## Audit checklist

| Area | What to check | How |
|------|---------------|-----|
| Indexation | noindex, canonical conflicts, `site:` coverage | fetch pages + web_search `site:domain` |
| robots.txt | blocked paths, sitemap line, crawl-delay | `web_fetch https://domain/robots.txt` |
| Sitemap | valid XML, fresh URLs, sample 404s | fetch `/sitemap.xml` (+ index children) |
| Redirects | chains, http→https, www, trailing slash | fetch variants; note status chain |
| Meta / head | title, description, robots, canonical, hreflang | sample 5-15 money pages |
| Structured data | JSON-LD presence/type | parse `<script type="application/ld+json">` |
| Internal links | orphans, deep money pages | sample nav + footer + in-content |
| Performance | LCP/CLS/INP hints only | ask for PSI/CrUX URL or note "requires PSI" |
| HTTPS | mixed content signals | fetch http upgrade + page assets notes |
| Hreflang | reciprocal FR/EN/AR | head tags on language variants |

## Workflow

1. Run `seo` action `pipeline` first for a full audit, or compose `crawl`, `audit`, `schema`, `psi`, `crux`, `score`, and `report`.
2. Treat engine findings and their evidence objects as the source of truth.
3. Use `serp_snapshot` only with DataForSEO or Semrush credentials and retain source/confidence.
4. Preserve every `data_gap`; do not replace missing PSI, CrUX, ranking, position, or volume data with estimates.
5. Add expert interpretation and prioritize findings as Critical / Important / Nice-to-have.
6. The canonical health score comes from `seo` action `score`. For legacy imported findings only, the helper remains available:

```bash
python navin/skills/seo-technical-auditor/scripts/audit_score.py findings.json
```

Or from workspace: `python <skill_dir>/scripts/audit_score.py seo/audit-findings.json`

7. Save report under `seo/technical-audit-<domain>-<date>.md` and feed `seo-report-*.html`.

## Finding severity

| Severity | Meaning | Examples |
|----------|---------|----------|
| Critical | Blocks crawl/index or money page | sitewide noindex, sitemap 404, https broken |
| Important | Material ranking/UX drag | canonical loops, missing titles, thin money pages |
| Nice to have | Polish | missing FAQ schema, minor meta length |

## Report format

```markdown
## Technical SEO Audit - <domain>
Date / scope / pages sampled

### Critical (blocks ranking)
1. issue - URL - evidence - fix - effort (S/M/L)

### Important
...

### Nice to have
...

### Quick wins (this week)
- ...

### Data gaps
- Core Web Vitals: requires PageSpeed Insights / CrUX
- Index coverage: prefer MCP `search-console` when connected; else GSC export if site: is inconclusive
```

## Rules

- Prefer live GSC via MCP `search-console` (Settings → MCP) for index coverage, queries, and page performance when available; otherwise ask for a CSV export or mark the gap.
- Every finding needs a URL + observed evidence (status, snippet, header).
- Never invent Lighthouse, CrUX, volume, or position data; use the engine's explicit `data_gap`.
- Prioritize by impact × effort; do not dump 100 undifferentiated issues.
- Sample-based audit is honest: state sample size, not "full crawl of N pages" unless you crawled them.
- Pair with `on-page-seo-optimizer` for page-level rewrites and `seo-monitoring` for tracking.

## Anti-patterns

- Auditing from memory without fetching the live site
- Claiming "Core Web Vitals fail" without data
- Mixing content strategy opinions into technical Critical