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 rumotion/ai-agent-project-template --skill human-voice-draftinggit clone --depth 1 https://github.com/rumotion/ai-agent-project-templateWrote 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/rumotion/ai-agent-project-template/human-voice-drafting)<a href="https://agentmods.dev/skills/rumotion/ai-agent-project-template/human-voice-drafting"><img src="https://agentmods.dev/badge/skills/rumotion/ai-agent-project-template/human-voice-drafting/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/rumotion/ai-agent-project-template/human-voice-drafting"><img src="https://agentmods.dev/badge/skills/rumotion/ai-agent-project-template/human-voice-drafting.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 | $0.00092 | $0.01799 |
| Opus 5.5 | $0.00037 | $0.00720 |
| Sonnet 5.5 | $0.00018 | $0.00360 |
| Haiku 4.5 | $0.00009 | $0.00180 |
Grade A, and why
human-voice-drafting 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 13d 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 — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Human Voice Drafting
Why this rule exists
This skill was codified after a real incident. A user sent an email drafted by an assistant to a senior colleague. A third party read it and told the user it "screams AI written". The flagged sentence was the opening:
"That's on me — I sent you five things I was worried about and never said what I wanted you to do with them. Let me try again."
The email was substantively good. It was accurate, well organised, and it said the right things. It still damaged the user's standing, because a reader who decides a message was machine-written stops reading it as the sender's own thinking and starts reading it as something the sender did not bother to write. That judgement is not recoverable within the same thread, and it is not repaired by the quality of the content.
This applies to every word that leaves the project under the user's name.
The hard rule
Anything the user will send under their own name must be written so that no recipient would suspect it was drafted by an AI. This outranks completeness, structure and polish. A shorter, plainer message that reads as human is always the correct trade against a thorough one that does not.
If a construction below is needed to say something important, say it another way. Do not keep the construction and hope it passes.
Banned constructions
These are the tells. None of them may appear in a draft.
1. The confessional opener and self-narration
Never open by narrating your own failure, apologising in a balanced clause, or announcing a retry. Never describe the message inside the message.
- ❌ "That's on me — I sent you five things and never said what I wanted. Let me try again."
- ❌ "Let me be clearer about what I'm asking."
- ❌ "I realise I wasn't specific enough in my last message."
- ✅ "Sorry, my last email wasn't clear about what I needed."
One short apology, then the point. A person writes six words of regret, not a sentence diagram of it.
2. Em-dashes doing dramatic work
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 13d ago First seen · 164 lines · 92 tokens per session scan A 4bc74cf42636
human-voice-drafting is a skill published in the GitHub repository rumotion/ai-agent-project-template (1 stars, last pushed 13d ago), licensed MIT. It adds 92 tokens to every session and 1,799 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-10-04.
Other skills, from other repositories
saas-builder
Clone, verify, map, and build on top of ixartz/SaaS-Boilerplate for a user's SaaS idea. Use when a user wants to reuse SaaS Boilerplate, evaluate how their product fits it, or build product-specific pages, database schema, roles, permissions, MVP features, and launch scope on top of the boilerplate.
dashboard-builder
Build self-contained interactive HTML dashboards with charts, filters, and tables. Generates a single browser-openable file — no server or dependencies required.
bionic-decision-engine
Unified mathematical arbitrator for all resource allocation decisions — money, time, energy, relationships. Absorbs 46 decision protocols + 24 strategy protocols into one dense engine.
trading-risk-gate
Unified pre-trade safety gate: Ruin check (Law #1), ergodicity audit, and win-rate dominance validation. Absorbs: ergodicity-check, law-of-ruin, win-rate-dominance.
skill-creator-primer
You MUST load this skill before the skill-creator skill AND before making ANY change to, or conducting a review of ANY Agent Skill. Triggers include creating, editing, reviewing, or contributing to any part of an Agent Skill (description, frontmatter, body, references, scripts, trigger evals, conflicts, etc).
database-patterns
DB schema design and query tuning: normalization, indexing, N+1, transactions, EXPLAIN. Triggers: schema, index, slow query, N+1, PostgreSQL, MySQL, EXPLAIN, deadlock, query plan.