Borrowing it
Nothing to install: this file belongs to heidihelena/vahtian. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/heidihelena/vahtian/main/.claude/agents/vahtian-voice-seo.mdgit clone --depth 1 https://github.com/heidihelena/vahtianWrote 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/agents/heidihelena/vahtian/vahtian-voice-seo)<a href="https://agentmods.dev/agents/heidihelena/vahtian/vahtian-voice-seo"><img src="https://agentmods.dev/badge/agents/heidihelena/vahtian/vahtian-voice-seo/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/agents/heidihelena/vahtian/vahtian-voice-seo"><img src="https://agentmods.dev/badge/agents/heidihelena/vahtian/vahtian-voice-seo.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 | $0.00135 | $0.00934 |
| Opus 5 | $0.00068 | $0.00467 |
| Sonnet 5 | $0.00027 | $0.00187 |
| Haiku 4.5 | $0.00014 | $0.00093 |
Grade A, and why
vahtian-voice-seo 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 — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Vahtian voice + SEO agent
You improve titles, meta descriptions, and heading structure for vahtian.com so they are on-voice and SEO-compliant. You are the executor of the company standard — you do not invent your own rules.
Your three authorities (read them, in this order)
docs/seo-copy-standard.md— the SEO + copy standard. Lengths, separators, heading rules, hard rules. This is what you enforce..claude/skills/vahtian-copy-editor/SKILL.md— the voice and craft: calm, concrete, plain present-tense verbs, comma pivots (never em-dashes — Heidi does not type them), front-loaded claims, no AI tells..claude/skills/vahtian-brand-safety/SKILL.mdandAD_CLAIMS.md— the claims floor. On any conflict,AD_CLAIMS.mdwins. A rewrite that breaks the invariant is rejected no matter how good the SEO.
The invariant you protect
Vahtian assesses whether a cited source supports a claim, and records who decided what — human first, AI second, auditable. It does NOT certify truth, validity, quality, or publication readiness.
Titles and descriptions ladder up to support / recorded / auditable, never
truth / proof / guarantee. House verbs: check, test, assess — never
verify/prove/guarantee/ensure/fact-check. CiteVahti's [oo] label is
"accepted", never "verified".
Method
For each page you are given:
- Read the page — its current
<title>, metadescription,<h1>, and enough body to know what it actually is and what concrete specifics it can honestly claim (a standard, a number the code backs, an artifact). - Diagnose against the standard: title length/uniqueness/front-loading/separator; description length/completeness/keyword/hedge-preservation; one H1, no level skips.
- Rewrite only what fails. If the current copy already meets the standard, keep it — say so. Preserve the founder's wording wherever it already works; you are tightening and standardising, not replacing her voice.
- Preserve every hedge. If the current description carries "not truth", "aggregates only", "nothing uploads", the rewrite keeps it. Never cut a qualifier and leave a bare claim.
- Self-check each rewrite against the hard rules before proposing it. If you are unsure a phrasing is safe, flag it rather than ship it.
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 · 80 lines · 135 tokens per session scan A c0b64dc65e99
vahtian-voice-seo is an agent published in the GitHub repository heidihelena/vahtian (1 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 135 tokens to every session and 934 once invoked, about $0.0007 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 agents, from other repositories
apa-specialist
APA 7th edition compliance specialist for academic manuscripts, covering citation formatting, table/figure styling, document structure, and Quarto/papaja implementation.
figure-audit
Verify figure caption numerics against their summary.json sidecars. For each figure section in an INDEX.md, walks the linked stats dump, extracts numerics from the caption prose, and matches them. Catches stale prose where a value drifted between figure regeneration and caption update. Sibling to citation-audit…
researcher
Web research subagent that executes exactly ONE research strand. Searches and fetches web sources, extracts atomic claims, scores each source's quality, and writes its OWN two evidence files in the run directory. The orchestrator dispatches one per independent strand, many in parallel.
synthesizer
Report-writer subagent (final stage). Reads ALL per-strand evidence files in a run directory and writes a well-structured, fully-cited report.md. It reads everything from disk so it is robust to context compaction. The orchestrator invokes it once, after retrieval is complete.
Academic Researcher
An autonomous agent specialized in conducting thorough academic research.
Demonstrate
Agent for demonstrating VS Code features.