Read
Content, metadata and page experience
Is there real, readable content, correct metadata, and a good page experience?
A reached page still needs substance: a real title, one meaningful h1, body copy that matches the intent, descriptive metadata and image alternatives. It also covers mobile and Core Web Vitals, which are genuine build-time fixes.
Paste into Claude Code, Cursor, Codex or any agent that can open a link. If yours cannot, use Copy full text.
Reach for it when you say
- “Missing meta description”
- “Duplicate titles”
- “My site is slow”
- “Is it mobile-friendly?”
Installed as a skill, it triggers on requests like these by itself.
- Step 01Diagnose content, metadata, semantics and page experience on the served HTML
- Step 02Fix the weakest first
- Step 03Verify on the rendered output
- Step 04Report what changed and why it matters
Three modes, three ways in
Run it in audit mode first: it records findings and changes nothing. Then fix applies only the findings you approve, and re-check re-tests them later. Switch modes on the prompt above.
- 01Paste the link promptOne line that points your agent at the skill on GitHub. Nothing to install, always the latest version.
- 02Paste the full textFor agents that cannot open links. The skill is self-contained, so this works in any chat.
- 03Install the packInstalled as skills, it triggers by itself when you describe the problem, and can hand work to the other skills.
The files
Questions about this skill?
Ask me on LinkedIn, or open an issue on GitHub if something is wrong or missing.