---
name: meta
description: "Writes a post's search-facing metadata: meta title, meta description, URL slug, lead image tags, and a minimal Article JSON-LD. Use when the user asks for the meta, an SEO title, a slug or permalink, og:image, article schema, what shows in search results before the click, or Google Discover thumbnails. For on-page keyword work, see seo-optimize."
metadata:
  version: 0.4.1
title: meta
canonical_url: https://skilld.dev/gh/rightblogger/bloggingskills/meta
last_updated: 2026-11-18T20:11:23.000Z
---

> Index of every page on this site: [llms.txt](https://skilld.dev/llms.txt)

> **Skill from skilld.dev.** Follow the user's requested action.
> If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it.
> Otherwise, follow the instructions below for this session. You do not need to install anything.
>
> If the user asked to install this Skill, run `npx skilld install rightblogger/bloggingskills/meta`. Install writes the Skill files into the project, so every session loads them.
>
> ## Fork workflow
>
> A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation.
> 1. Check `./skills/meta`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target.
> 2. Read [source metadata](https://skilld.dev/api/v1/skills/rightblogger/bloggingskills/meta) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit.
> 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call:
>
> ```sh
> git init --quiet "$temporary_dir"
> git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit"
> git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD
> ```
>
> Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop.
> 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/meta`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md.
> Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical.
> 5. In the project root, run `skilld install ./skills/meta --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/meta --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure.
> 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks.

# Meta

You write the three fields Google shows before anyone clicks: the meta title, the
meta description, and the URL permalink. These are public the moment the page is
indexed, so they are the worst place for AI-ese to leak. One stray em dash or one
"comprehensive guide" here is seen by every searcher, including the ones who never
scroll. AI assistants also quote titles and descriptions verbatim, which raises the
stakes on what they say.

You're done when each field fits its cap, leads with the searcher's words, states
something the post actually delivers, and passes the rulebook's hard rules.

## Initial assessment

- If `.agents/blog-context.md` exists, read it. A captured author voice and the
  site's URL conventions override the defaults below.
- Read the hard rules and the Titles section of the house-style rulebook at
  `../house-style/references/ai-tells.md`. Every field has to pass them.
- Get the primary keyword. If the user did not give one, read it off the post's H1
  and first paragraph and say which one you used. If Search Console data is
  available, use it to pick the exact phrase people search rather than guessing.
- Skim the post (or topic) so the description matches what the page delivers. A
  promise the post does not keep is a click that bounces.

## Rules

Lead with the proven keyword the reader actually types, near the front of the
title where it survives truncation. Keep the title close to the post's H1 and to
the query. Search engines rewrite titles that drift from the page, and they rewrite
most meta descriptions anyway, so a description that matches the page is the one
most likely to survive.

Carry a fact from the post instead of teasing it. "Disabling emoji scripts removes
three HTTP requests" outlives "Find out how to speed up your site," because people
and AI assistants can use it as is. The fact has to come from the post body: never
invent a result, ranking, or number, and never write "our tests" or "we found"
unless the post says so. If the post makes no specific claim, describe concretely
what it covers.

When a field holds alternatives (two prices, several options, a range), put the
value the site most wants quoted first. Skimmers and AI answers take the first
number or name they see. Write plain language a 12-year-old would follow.

Metadata attracts its own filler on top of the rulebook's title molds: "comprehensive
guide," "discover," "learn how." Both sets apply to every field, not only the title.
Say what the page gives the reader.

No em or en dashes in any field; a comma or colon does the same job. A hyphen inside
a compound word is fine.

## Field caps

- Meta title: 60 characters maximum, aim for 50 to 58 so it has headroom. Keyword
  near the front. Over 60 gets truncated in the SERP.
- Meta description: 155 characters maximum, aim for 135 to 150. Active voice, lead
  with the fact or benefit, work the keyword in naturally.
- Permalink: 2 to 5 natural-language words, lowercase, hyphen-separated, readable as
  a phrase. No stop-word padding and no IDs. Strip years and numbers unless they are
  part of the topic. Never change the slug of a published post unless the
  user asks; that needs a redirect.

Show the character count beside the title and the description so the user can see
the caps hold. Models miscount characters, so if you can run code, count with it
(for example `len("...")`); otherwise count carefully and leave a few characters of
headroom.

## Image and structured data

Google Discover and AI answers show a thumbnail, and Google picks it from `og:image`
and the page's image markup. Discover demotes clickbait and favors original,
in-depth posts, so the title rules above apply to `og:title` too.

- `og:image`: the post's lead image, at least 1200px wide. The author's own photo or
  screenshot beats stock, since Discover rewards original work. Use only a real image
  path from the user or the post; with none, say the post needs a lead image instead
  of inventing a URL.
- `max-image-preview:large` in the robots meta tag, so Discover can show the large
  card. Many sites set it site-wide already; check before adding it.
- Article JSON-LD with `headline`, `image`, `datePublished`, `dateModified`, and
  `author` as a Person with `name` and `url` pointing to a real author page. Take
  every value from the post or `.agents/blog-context.md` and leave out any field you
  can't fill truthfully. Change `dateModified` only when the content changed in
  substance (see refresh).

This markup earns thumbnails and ties the post to its author. It doesn't earn AI
citations on its own (see ai-seo), so keep it to these fields. If the CMS or an SEO
plugin already outputs them, give the values to fill in rather than raw markup.

## Output

Return the three fields, labeled:

- Meta title, with its character count.
- Meta description, with its character count.
- Permalink.
- The lead image recommendation and the Article JSON-LD block, unless the user only
  asked for the title and description.

Offer a couple of title or description variants when a different angle on the
keyword could win more clicks. Keep every variant inside the caps.

If the post itself gets in the way (an outdated year in the H1, a claim it
contradicts elsewhere), write the fields from what's true and add one line telling
the author what to fix in the post.

## Related skills

article · humanize · seo-optimize · blog-context · house-style · ai-seo
