titles-meta
Skill0 starsChanged 3 days agoMITView on GitHub
Rewrite page titles and meta descriptions using your own Search Console data. Finds under-clicked queries with your own CTR curve (ctr), ranks positions 4-20 with real impressions (striking), checks the live title and description (page), applies Google's title link and snippet guidance, supplies templates, and measures each rewrite against untouched control pages with measure-rollback. Use when the user says 'improve CTR', 'rewrite my titles', 'meta descriptions', 'low click-through', 'title tag test', 'pages with impressions but no clicks', 'striking distance'.
In this repository44 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-readySkill
- ai-citationsSkill
- ai-crawlersSkill
- ai-logsSkill
- ai-trackerSkill
- backlinksSkill
- cannibalizationSkill
- canonicalsSkill
- citabilitySkill
- comparison-pagesSkill
- competitorsSkill
- content-briefSkill
- content-decaySkill
- content-qualitySkill
- core-web-vitalsSkill
- ecommerce-seoSkill
- ga4-aiSkill
- ghost-publishSkill
- gsc-bingSkill
- hreflangSkill
- imagesSkill
- indexingSkill
- internal-linksSkill
- llms-txtSkill
- local-seoSkill
- measure-rollbackSkill
- onpageSkill
- platformsSkill
- pseo-quality-gateSkill
- quick-winsSkill
- redirects-migrationSkill
- schemaSkill
- seo-planSkill
- seo-reportSkill
- setupSkill
- site-architectureSkill
- sitemapsSkill
- static-site-publishSkill
- sxoSkill
- tech-auditSkill
- topic-clustersSkill
- weekly-checkSkill
- wordpress-publishSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Search Console data for at least 28 days, same date range for every file: the API (gsc, see gsc-bing) or the Performance report exports.
- Distinctive: line 29
Add striking-distance rows (positions 4-20, at least 100 impressions):
- Distinctive: line 39
Write the new title and description with the rules and templates below.
- Distinctive: line 56
Put a year or date in a title only if the page is updated on a schedule: Google lists obsolete dates as a rewrite cause.
- Very specific: line 59
Use the measure-rollback skill: snapshot before, change, snapshot after, choose control pages of the same type and similar position that …
- Distinctive: line 72
Do not promise Google will use your title or description.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Titles and descriptions from query data
Goal: pick the pages where a better title or description can earn clicks the page already has impressions for, rewrite them with Google's documented guidance, and measure the result. Google's traffic-drop guidance says that when only clicks fall while impressions hold, improve titles and snippets; if impressions fell too, the cause is elsewhere (see content-decay).
Inputs
- Search Console data for at least 28 days, same date range for every file: the API (
gsc, seegsc-bing) or the Performance report exports. - Pages the user owns and may edit.
Steps
- Get data:
npx sitetiller gsc example.com --start 2026-08-12 --end 2026-11-09 --dims page --out pages.csv npx sitetiller gsc example.com --start 2026-08-12 --end 2026-11-09 --dims query --out queries.csv npx sitetiller gsc example.com --start 2026-08-12 --end 2026-11-09 --dims query,page --out query-page.csv - Build your own CTR curve and find under-clicked rows:
Output:npx sitetiller ctr pages.csv --jsoncurve(position 1-10 mapped to{ctr, rows}: the median CTR of rows with at least 50 impressions, kept only for positions with at least 3 rows) andlowCtr(rows with at least 100 impressions whose CTR is below 60% of your median at their position, withexpectedCtrandmissedClicks). This is your site's curve, not an industry average. If a position is missing fromcurve, do not fill it in from memory. - Add striking-distance rows (positions 4-20, at least 100 impressions):
Output per row:npx sitetiller striking queries.csv --jsonquery,clicks,impressions,ctr,position,opportunity. For these, a title change rarely moves position; content and links matter more (seeonpage,internal-links). - For each candidate page, list its top queries from
query-page.csvand read the live elements:
Usenpx sitetiller page https://example.com/guides/invoice-template --jsontitle,description,h1. Search the top query in a logged-out window and note the title and snippet Google actually shows. If the title differs from yours, check the six rewrite causes in theonpageskill first: a rewritten title link usually means the page is inconsistent, so fix the page, not only the tag. - Write the new title and description with the rules and templates below. Give two options per page and say what each tests.
- Measure (below). Change one thing per page per window.
Rules from Google's documentation
- Titles: descriptive and concise; distinct on every page; no repeated words or keyword stuffing; brand at the start or end behind a delimiter; same language as the page; the first visible H1 should agree. No length limit exists; display is cut to fit the device width, so put the distinguishing words first.
- Descriptions: unique per page; summarise the page accurately and specifically; may include facts such as author, date or price; no keyword lists. Google may build the snippet from page text instead, and can show different snippets per query.
Templates
Fill every slot with something true on the page. If a slot has nothing true, drop it.
| Page type | Title | Description |
| How-to / guide | {Task} {object}: {method or number of steps} | {What the reader gets}. {One specific fact, example or number from the page}. {Who wrote or tested it, or date}. |
| Template / tool | Free {thing} {format}: {what it does} | {What it produces} in {time or steps}. {Formats, limits, price if any}. |
| Comparison | {A} vs {B}: {the deciding difference} | {Who each one fits}. Compared on {criteria}, checked {date}. {Disclosure if you are one of them}. |
| Pricing / product | {Product} pricing: {plans and price range} | {Plans, free tier or trial terms}. Prices checked {date}. |
| Data / research | {Metric}: {headline finding} | {Method in 10 words}. {Sample size}. {Date range}. |
| Brand suffix: ` | Brand` when the brand adds trust; omit it where space matters. | |
| Put a year or date in a title only if the page is updated on a schedule: Google lists obsolete dates as a rewrite cause. |
Measure each rewrite
Use the measure-rollback skill: snapshot before, change, snapshot after, choose control pages of the same type and similar position that you do not touch, wait 14 to 28 days, then:
npx sitetiller measure --daily clicks.csv --jsonJudge a title change on clicks and CTR at stable impressions and position. Do not read a result from a few days of data.
Output to the user
- Table: page, top query, position, impressions, CTR, your median CTR at that position, missed clicks (from
ctr). - Per page: current title and description (quoted), what Google shows if different, two rewrites, the hypothesis each tests.
- Control pages chosen and the measurement window.
Guardrails
missedClicksis an estimate from your own median, not a forecast. Do not promise a CTR lift.- Do not promise Google will use your title or description.
- Split branded and non-branded queries first (
npx sitetiller brand queries.csv --terms yourbrand); brand queries distort the curve. - Change only titles and descriptions in the window, or the measurement is void.
- Edit only pages the user owns and named.
References
references/sources.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

