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 ur-grue/autopunk-media-skills --skill ai-writing-detoxgit clone --depth 1 https://github.com/ur-grue/autopunk-media-skillsWrote 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/ur-grue/autopunk-media-skills/ai-writing-detox)<a href="https://agentmods.dev/skills/ur-grue/autopunk-media-skills/ai-writing-detox"><img src="https://agentmods.dev/badge/skills/ur-grue/autopunk-media-skills/ai-writing-detox/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/ur-grue/autopunk-media-skills/ai-writing-detox"><img src="https://agentmods.dev/badge/skills/ur-grue/autopunk-media-skills/ai-writing-detox.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.03820 |
| Opus 5 | $0.00034 | $0.01910 |
| Sonnet 5 | $0.00014 | $0.00764 |
| Haiku 4.5 | $0.00007 | $0.00382 |
Grade A, and why
ai-writing-detox 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 — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI Writing Detox
What This Skill Does
Rewrites AI-flavoured copy into a publishable register by stripping the language tells of an LLM draft — buzzwords, throat-clearing, false-inclusive openers, and the "not just X — Y" rhetorical tic — and supplies the canonical banned-list that runtime hooks read to flag drafts before they ship.
When To Use This Skill
- A reporter or producer suspects an LLM drafted (or heavily assisted) the copy and wants a clean rewrite
- A subeditor is reviewing a piece that reads "fluently bland" — grammatically correct, recognisably synthetic
- Before posting any external surface (article, show notes, newsletter, README) where the audience is media professionals who notice these tells
- When onboarding a new contributor whose first drafts default to AI register
- As the source of truth for the
ai-slop-detectorruntime hook (see Hook Contract below)
What You Need To Provide
Required: The draft text you want detoxed. Optional: Outlet or surface (news article, show notes, internal memo, README), audience register (general public, working journalists, investor-facing), any preserved terms (e.g., a term that looks like jargon but has been defined earlier in the piece), word-count constraint if the rewrite must hit a length.
How the Assistant Approaches This
- Reads the full draft and identifies every hit against the Hard ban list — these are non-negotiable rewrites.
- Reads it again for Soft warn patterns — surfaces them with one-line reasons; the editor decides whether to keep, soften, or cut.
- Produces a rewritten draft that preserves every load-bearing fact and quote, swaps banned vocabulary for the plainest accurate alternative, and dismantles "not just X — Y" / "it's not about X, it's about Y" constructions into direct claims.
- Returns a short before/after table for the most consequential changes so the writer can learn the pattern, not just take the rewrite.
Output Format
- A clean rewritten draft as the headline deliverable, ready to drop in.
- A
Changestable (3–8 rows): original phrase → replacement → which lens motivated the cut (Plain-language default,Cut the meta,Honest hedging,Concrete over abstract,Reader respect). - A
Soft-warn flagslist for items that were left in but the editor should review. - Tone is direct and practical — written for an editor who will sign off the final cut.
- Length scales with the input. The rewrite is at most the length of the original; usually shorter once the buzzwords come out.
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.
- 12d ago First seen · 201 lines · 69 tokens per session scan A ed65f74937fc
ai-writing-detox is a skill published in the GitHub repository ur-grue/autopunk-media-skills (32 stars, last pushed 12d ago), licensed MIT. It adds 69 tokens to every session and 3,820 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 2026-08-30.
Other skills, from other repositories
zero-slop
Edit drafts into natural prose, inspect AI-sounding patterns, or review how a specified audience might respond passage by passage. Zero Slop runs inside the user's existing AI assistant with local tools that protect source details. Use for humanizing or de-slopping writing, polishing outward-facing prose, social…
clarity-and-concision
Teaches techniques for achieving relentless clarity and ruthless concision in business writing, including the 20-word sentence rule, concrete detail over abstraction, word economy tactics, and the reader's perspective test. Features extensive before/after examples demonstrating how bloated corporate prose can be cut…
voice-update
Update context/voice-and-style.md or context/about-me.md from one of three sources. Manual (user dictates a single new rule, sample, or career fact). Memory (batch pull from Claude Code's auto-memory in /.claude/projects/ /memory/). Sent-mail (analyze the last 20-50 sent Gmail messages and propose updates from…
yourself-story
Drafts and critiques bios, LinkedIn About sections, intro slides, "tell me about yourself" interview answers, and personal-essay openers using Adam Bryant's 500-CEO research and Lauren Weinstein's warmth+competence frame. Different from pitch-coach (which is product-shaped) and cold-email-coach (which is…
compression
Cuts a draft to a target word count without losing substance. Two modes behind one skill. target-count hits a specific number (200 for a cold email, 500 for an op-ed, 6 for a product summary). redundancy catches the specific failure where a phrase says what the verb or context already implied — "going forward" after a…
style-tells
Scrubs the three surface tells that make prose feel AI-generated, padded, or jargon-heavy. Three targets behind one skill, picked via the --target arg. em-dashes (default off in cold email / memo / Slack, max one per page in op-eds), adverbs (the -ly and intensifier pile), jargon (the Silicon Valley / consultant…