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 larsboes/Axon --skill human-writinggit clone --depth 1 https://github.com/larsboes/AxonWrote 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/larsboes/axon/human-writing)<a href="https://agentmods.dev/skills/larsboes/axon/human-writing"><img src="https://agentmods.dev/badge/skills/larsboes/axon/human-writing/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/larsboes/axon/human-writing"><img src="https://agentmods.dev/badge/skills/larsboes/axon/human-writing.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 | $0.00232 | $0.03219 |
| Opus 5 | $0.00116 | $0.01610 |
| Sonnet 5 | $0.00046 | $0.00644 |
| Haiku 4.5 | $0.00023 | $0.00322 |
Grade A, and why
human-writing 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 8d 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 — 226 lines — stays where its author put it; the contents beside it link to each section on GitHub.
human-writing
Read this first, because the most common misunderstanding sinks the whole thing.
This skill does not supply taste, and it has no house style. It does two narrow things. It removes the specific cues that make text read as machine-written, and where the text reads as AI because nobody ever chose a voice, it forces a deliberate one. Argument and judgment stay with the author. A guardrail is not a writer.
Most of the work is mechanical, and the plain fix is usually correct. Nearly every cited
tell has a fix that needs no taste at all: the em dash becomes a comma or a period, delve into becomes look at, the sycophantic opener gets cut, the as an AI boilerplate gets
deleted, the "in conclusion" recap goes. Lead with the plain fix. Reserve the voice work for
the genuinely stylistic calls, and for prose that reads as empty because nobody decided what
it was.
The trap to avoid (this is the whole point)
Every anti-AI-writing effort fails the same way: it replaces one default register with another. The 2024 tell was the smooth corporate voice. The 2026 over-correction is its mirror image, the "trying not to sound like AI" voice: staccato fragments, forced lowercase, a "here's the thing" cold open, a swear dropped in to seem casual, and the truly desperate move of pasting fake typos to beat a detector. Readers clock the second one just as fast.
Treat the over-corrected register as its own tell. Three moves in particular. Conspicuous em-dash avoidance: the fix for a dash is a comma or a period, not an ellipsis and not a sentence visibly contorted around the gap, because the bending is as legible as the dash was. Manufactured casualness: a "honestly", a lowercase "i", a "lol" bolted onto otherwise formal writing reads as costume. And rhythm that is uniform in a new way: all-short sentences are as mechanical as all-medium ones. Evenness is the tell, whichever length it settles on.
A tell is an unspecified default, not a banned word. If the writing genuinely calls for a formal register, or the author genuinely loves the em dash, that is not slop. Honor the escape hatches (see The scanner). Full catalog: references/over-correction.md.
What ships with it
29 files 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.
- references/anti-hallucination.md 4.4 KB
- references/cited-vs-matched.md 3.9 KB
- references/discourse-and-structure.md 4.2 KB
- references/maintenance.md 2.9 KB
- references/over-correction.md 3.2 KB
- references/registers.md 3.9 KB
- references/structural-craft.md 7.1 KB
- references/tells.md 24 KB
- references/types.md 4.4 KB
- references/voice-template.md 3.6 KB
- references/writing-with-intent.md 4.8 KB
- scripts/ai_prose_patterns.json 30 KB
- scripts/detect_ai_prose.py 1.6 KB runs code
- scripts/human_voice_linter/__init__.py 1.8 KB runs code
- scripts/human_voice_linter/analyze.py 7.7 KB runs code
- scripts/human_voice_linter/api.py 1.3 KB runs code
- scripts/human_voice_linter/autofix.py 6.1 KB runs code
- scripts/human_voice_linter/checks.py 39 KB runs code
- scripts/human_voice_linter/cli.py 8.6 KB runs code
- scripts/human_voice_linter/config.py 3.4 KB runs code
- scripts/human_voice_linter/defaults.py 4.8 KB runs code
- scripts/human_voice_linter/directives.py 3.1 KB runs code
- scripts/human_voice_linter/hit.py 1.6 KB runs code
- scripts/human_voice_linter/patterns.py 2.6 KB runs code
- scripts/human_voice_linter/report.py 4.0 KB runs code
- scripts/human_voice_linter/schema.py 6.1 KB runs code
- scripts/human_voice_linter/score.py 2.8 KB runs code
- scripts/human_voice_linter/textutil.py 12 KB runs code
- scripts/human_voice_linter/util.py 399 B runs code
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.
- 8d ago First seen · 226 lines · 232 tokens per session scan A 077f02e34bab
human-writing is a skill published in the GitHub repository larsboes/Axon (1 stars, last pushed today), licensed MIT. It adds 232 tokens to every session and 3,219 once invoked, about $0.0012 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
orbit-notion
Open Orbit briefing skill — selected by the Orbit pipeline when Notion is the user's only connected connector, or when the user explicitly scopes their daily digest to Notion. Pulls the past 24 hours of document edits, comments, mentions, and database row changes from the user's authenticated Notion connection and…
motion-frames
A single-frame motion-design composition with looping CSS animations — rotating type ring, animated globe, ticking timer, parallax labels. Renders as a hero video poster you can hand straight to HyperFrames or any keyframe-based exporter. Use when the brief asks for "motion design", "animated hero", "loop", "video…
hr-onboarding
A new-hire onboarding plan as a single page — first week schedule, buddy + manager intro, learning track, equipment checklist, and "you're set when…" outcomes. Use when the brief mentions "onboarding", "new hire", "first week plan", or "入职".
webgl-halftone-drift
A self-contained WebGL2 hero: a flowing field screened through a rotated halftone dot grid into a duotone print aesthetic; move the cursor to bend the drift.
frame-light-leak-cinema
Film light leaks, grain, 16:9 letterbox, and large serif type for cinematic openings or chapter cards.
card-xiaohongshu
Xiaohongshu-style knowledge cards, arranged as a swipeable multi-card carousel.