| name | seo-onpage |
| description | Shopify on-page SEO audit sub-skill. Scores title tags, meta descriptions, H1 headings, image alt text, and internal linking per page. Use when user runs /seo onpage or when seo-audit delegates on-page checks.
|
SEO On-Page Audit
On-page SEO sub-skill for Shopify stores. Scores title tags, meta descriptions, H1 headings,
image alt text coverage, and internal linking per page. Aggregates results by check across pages.
Category weight: On-Page = 25% (0.25) of overall SEO Health Score.
Quick Reference
| Command | What it does |
|---|
/seo onpage <url> | Run on-page SEO checks against a Shopify store |
| (via seo-audit) | seo-audit spawns audit-onpage agent directly — this sub-skill is not used |
Context Intake
This skill is invoked directly via /seo onpage <url>.
When invoked directly:
- If store URL not provided, read
~/.claude/skills/seo/profiles/karved.json and use preset
- Run
python3 ~/.claude/skills/seo/scripts/crawl_store.py <url> to generate crawl data before proceeding
- Save output as
seo-audit-crawl.json in the current directory
- Then proceed to Execution Logic
Note: When invoked by seo-audit, this sub-skill is NOT used — seo-audit spawns the audit-onpage agent directly.
Execution Logic
- If invoked directly (not from seo-audit):
a. If store URL not provided, read
~/.claude/skills/seo/profiles/karved.json and use preset
b. Run python3 ~/.claude/skills/seo/scripts/crawl_store.py <url> to generate seo-audit-crawl.json
- Spawn agent
audit-onpage via Task tool with prompt:
"Run on-page SEO audit on {store_url}. Crawl data at {cwd}/seo-audit-crawl.json"
- Agent writes
seo-onpage-findings.md — present results to user
All check definitions, scoring logic, and output format are defined in the agent.
If invoked by seo-audit, this sub-skill is NOT used — seo-audit spawns the agent directly.
Quality Gates
This skill inherits all quality gates from seo/SKILL.md. Do NOT redefine or soften them. The audit-onpage agent applies severity levels locked by the quality gates.
Key inherited gates (for reference, not redefinition):
- Duplicate title tags = Critical
- Missing H1 = Critical
- Missing meta description = High
- Missing image alt text > 20% = High
Reference Files
Load these on-demand as needed — do NOT load all at startup.
Path resolution: All references are installed at ~/.claude/skills/seo/references/.
references/scoring-system.md — Weighted scoring algorithm, severity multipliers (Critical=5.0, High=3.0, Medium=1.5, Low=0.5), grade thresholdsreferences/benchmarks.md — Content Benchmarks table: title tag 50-60 chars, meta description 120-160 chars, image alt text coverage 80% minimum, internal links per page 3 minimum
The agent loads its own references during execution.
