Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add charlie947/voiceprint --skill skillgit clone --depth 1 https://github.com/charlie947/voiceprintWrote 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/charlie947/voiceprint/skill)<a href="https://agentmods.dev/skills/charlie947/voiceprint/skill"><img src="https://agentmods.dev/badge/skills/charlie947/voiceprint/skill/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/charlie947/voiceprint/skill"><img src="https://agentmods.dev/badge/skills/charlie947/voiceprint/skill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Output Handling · line 64 Output size or generation rate is not bounded. Unbounded output enables denial-of-service through resource exhaustion, log flooding, or context-window stuffing.Fix: Set explicit limits on output length, generation count, and rate. Use max_tokens and truncation to prevent unbounded output.
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 | $0.00083 | $0.01650 |
| Opus 5 | $0.00042 | $0.00825 |
| Sonnet 5 | $0.00017 | $0.00330 |
| Haiku 4.5 | $0.00008 | $0.00165 |
Grade A, and why
voiceprint 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 9d 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 — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.
voiceprint
Turns someone's own spoken words into a finished article.
You do the research, the structure, the headings, the fact-checks and the polish. You do not write the sentences. They already did that, out loud.
The rule, and why it is the whole skill
Text reads as human in proportion to how much of the person's literal wording survives into it. Not their ideas. Their sentences.
Measured on one article, same source recording, varying only the proportion of model-composed sentences:
| Their own words | Verdict |
|---|---|
| 100% | 100% human |
| 87.5% | 100% human |
| 71.7% | 58% human |
| 61.7% | 57% human |
| 13.5% | 100% AI |
It is a threshold, not a slope. Past roughly 85% the verdict flips, and adding more composed text after that barely changes it.
The same article written from the same video by a model, using the person's facts and opinions, scored 100% AI. Sourcing the ideas from a human does not help. Only the sentences do.
So: about one sentence in eight may be yours. That is enough for joins and transitions. It is not enough for a section.
What this skill will not do
It will not make model-written prose read as human. Style profiles, humaniser passes and register-matching have all been measured and they all make it worse, not better. If asked for one, say so once and offer this instead.
It is not a detection-evasion tool. It produces writing that genuinely is the author's, because they said it.
Step 1 — Get their words
They need roughly 1.3x their target word count in spoken material. About 75% of what someone says survives the cut. A 1,500-word article needs around 2,000 spoken words, which is 11 to 13 minutes of talking.
Three ways in. Offer whichever fits.
A. Record now. Best quality. Ask them to talk through the piece as if explaining it to one person. Give them the section list first so they cover everything, then get out of the way.
B. Paste a transcript. They already have a video, podcast, voice note or webinar on the topic. Any of it works.
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.
- 9d ago First seen · 174 lines · 83 tokens per session scan A bc0d77a49294
voiceprint is a skill published in the GitHub repository charlie947/voiceprint (61 stars, last pushed 1mo ago), licensed MIT. It adds 83 tokens to every session and 1,650 once invoked, about $0.0004 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 2026-08-30.
Other skills, from other repositories
humanizer
Detects 55 AI writing patterns and rewrites text in five voice profiles so it reads like a specific human wrote it, with an optional 0-100 AI-tell score. Use when text sounds AI-generated or like a chatbot, when preparing a blog post, README, or LinkedIn post for publication, when auditing prose for AI tells, or when…
patina
Detect and rewrite AI writing patterns in Korean, English, Chinese, and Japanese text so it reads as if a human wrote it. Meaning-preservation (MPS) verified.
patina-aside
Use Patina's local options page and verified CLI rewrite when drafting or revising blog posts in Aside, including automatic blog workflows.
clarity
Draft, rewrite, or review reader-facing prose so it is specific, useful, and recognizably the author's without inventing facts or performing humanness. Use for essays, articles, newsletters, documentation, talks, launch copy, and other important prose that feels generic, hollow, or AI-shaped. Supports co-write…
book-genesis-full
Production Mode — full 17-phase pipeline with all 19 skills. Use when you want the complete industrial pipeline with research, evaluation, editorial packaging. For most users, use /book-genesis (Craft Mode) instead.
hook-craft
Specializes in chapter openings (hooks) and chapter endings (pulls). Every chapter must start with a reason to keep reading and end with a reason to turn the page. Runs after prose-craft, before chaos-engine. The skill that prevents the reader from putting the book down.