Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/Smolevich/pet-project-atlasnpx agentmods add skills/smolevich/pet-project-atlas/voiceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/smolevich/pet-project-atlas/voice)<a href="https://agentmods.dev/skills/smolevich/pet-project-atlas/voice"><img src="https://agentmods.dev/badge/skills/smolevich/pet-project-atlas/voice/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/smolevich/pet-project-atlas/voice"><img src="https://agentmods.dev/badge/skills/smolevich/pet-project-atlas/voice.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00069 | $0.00851 |
| Opus 5 | $0.00034 | $0.00426 |
| Sonnet 5 | $0.00014 | $0.00170 |
| Haiku 4.5 | $0.00007 | $0.00085 |
Grade A, and why
voice scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/atlas:voice
Run the repository's own linter against a page, then fix what it reports against STYLE.md.
The rules live in STYLE.md at the repository root. Read it. Do not restate its rules here or in your
answer — quote the section the linter names and fix against that section.
Before you touch anything
- Confirm you are inside the
pet-project-atlasrepository.scripts/lint-voice.mjsandSTYLE.mdboth have to exist. If they do not, stop and say so — this skill only works here. - Read the target page's frontmatter.
voice: guestmeans stop. A guest case keeps its author's voice: the ban list, the page shape and the length rules do not apply to it. Report that the page is a guest case and change nothing in its prose. The number rule still applies to guest pages, so an unsourced-number error on a guest page is a real error — fix that one by adding a source or deleting the number, and nothing else.
Run the linter
The script takes either a directory or a single file. Point it straight at the page you are fixing:
node scripts/lint-voice.mjs src/content/docs/<section>/<page>.md
For the whole repository, use the package script:
npm run lint:voice
Output shape, one finding per line:
<file>:<line>: <what broke>. <what to do>. (STYLE.md §N Section)
Errors exit non-zero and fail the build. Warnings keep the build green.
Fix each finding
Work one finding at a time, in the order the linter printed them. For each one open the section of
STYLE.md named in the parentheses and fix against it.
- §3 Ban list — a banned word is a symptom of a sentence with no claim in it. Replace the sentence with a concrete claim from the author's own log. A synonym from the thesaurus is not a fix.
- §2 Page shape — the four required H2 headings are matched literally, at level two, before any other H2. A heading with nothing under it counts as missing.
- §4 Numbers — either add the date to
updated:and the source tosources:, or delete the number. A URL goes in asTitle — https://…, and the title says why that link is there. Never soften a number into words. If you cannot find the source, the number goes. - §5 Length — a warning, not an error. Split the sentence. If a page throws many of these, say so: it usually needs cutting, not exceptions.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 82 lines · 69 tokens per session scan A 9478b578d44c
voice is a skill published in the GitHub repository Smolevich/pet-project-atlas (2 stars, last pushed 2d ago), licensed MIT. It adds 69 tokens to every session and 851 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
geo-visibility
Get cited and recommended by AI engines (ChatGPT, AI Overviews and AI Mode, Perplexity, Claude, Gemini). Input: a page or piece of content. Output: passage-level citability fixes (answer-first H2 blocks, self-contained chunks, definitions, sourced stats, comparison tables), a 5-pillar GEO score (0-100), an AI-crawler…
seo-content-collection-page
Optimize e-commerce collection, category, and product listing pages (PLPs) for Google and AI assistants. Input: a collection or category page (Shopify, WooCommerce, Magento, BigCommerce, PrestaShop, or custom). Output: a bottom-of-page SEO text block, faceted-navigation and filter URL control, pagination canonicals…
geo-tracking
Measure AI visibility without paid tools or API keys. Input: your site (GA4 and server logs) and a buyer prompt panel. Output: GA4 AI-traffic reporting (custom channel group plus referrer regex above Referral), monthly brand mention rate, citation rate, and share of voice versus competitors across ChatGPT, Perplexity…
seo-content-blog
Write blog articles that rank on Google and get cited by AI engines (ChatGPT, Perplexity, AI Overviews). Input: a keyword, topic, or existing draft. Output: a publish-ready article, outline, or brief built on a 12-element answer-first skeleton (question H2s, expert quotes, stats, FAQ, internal links, SERP-benchmarked…
seo-content-product-page
Optimize e-commerce product pages (PDPs) for Google and for AI assistants that now recommend products directly. Input: a product page or description (Shopify, WooCommerce, Magento, BigCommerce, PrestaShop, Wix, Webflow, or custom). Output: a rewritten PDP with unique copy, FAQ and definition blocks, review and…
seo-internal-linking
Design internal linking so authority flows to the pages that sell and every page stays crawlable. Input: a sitemap, an article, or a set of posts. Output: money-page mapping, orphan-page fixes, content silos and hub-and-spoke clusters, anchor-text variation, breadcrumb/menu/footer roles, and keyword cannibalization…