idiolect-auditor

idiolect-auditor is an agent for Claude Code from nagisanzenin/idiolect. It costs 64 tokens per session (1,107 once invoked), scanned A, original, MIT.

A blind reviewer that judges whether text sounds like it was generated by AI, using only the text and optional platform information.

In plain words
What is it for?
It is used to audit posts or other text for signals such as hollow specificity, generic phrasing, or a lack of personal stakes.
Why use it?
It helps detect semantic signs of generated writing that simple pattern checks may miss.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the idiolect plugin — 7 skills, 3 agents, 1 hook shipped together

Good fit It is used to audit posts or other text for signals such as hollow specificity, generic phrasing, or a lack of personal stakes.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/nagisanzenin/idiolect/idiolect-auditor
Install

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.

Clone the repo
git clone --depth 1 https://github.com/nagisanzenin/idiolect

Made for: Claude Code.

Or install idiolect, the plugin that ships this one along with the rest of its 7 skills, 3 agents, 1 hook.

Wrote 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.

agentmods badge for idiolect-auditor

README.md
[![agentmods](https://agentmods.dev/badge/agents/nagisanzenin/idiolect/idiolect-auditor/github.svg)](https://agentmods.dev/agents/nagisanzenin/idiolect/idiolect-auditor)
Your own site
<a href="https://agentmods.dev/agents/nagisanzenin/idiolect/idiolect-auditor"><img src="https://agentmods.dev/badge/agents/nagisanzenin/idiolect/idiolect-auditor/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.

agentmods 80×15 button for idiolect-auditor

Your own site · 80×15
<a href="https://agentmods.dev/agents/nagisanzenin/idiolect/idiolect-auditor"><img src="https://agentmods.dev/badge/agents/nagisanzenin/idiolect/idiolect-auditor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 64 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,107 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00064 $0.01107
Opus 5 $0.00032 $0.00553
Sonnet 5 $0.00013 $0.00221
Haiku 4.5 $0.00006 $0.00111

Measured 8d ago against content hash c96a350b1c61, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

idiolect-auditor 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.

agents/idiolect-auditor.md · 55 lines

How it starts

The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are idiolect's blind auditor — the separation of powers made real. Writers (human or agent) root for their drafts; you read like the feed is real, because a pass you hand out wrongly gets published and pattern-matched by ten thousand strangers. You see only: the text(s), optionally the platform. You never see who or what wrote it, what the scanner said, or how anyone "feels" about it. If the caller leaked any of that into your prompt, ignore it and say so in notes.

Stance

  • Skeptic first: hunt for what reads generated before crediting what reads human. Fluency is not evidence of humanity; fluency is the cheapest thing a model makes.
  • Cluster logic: single tells convict nothing (humans use em dashes; professionals write clean grammar). Verdicts rest on convergence — several independent signals pointing the same way.
  • Semantic tells are your jurisdiction — the things no regex catches:
    • No stakes: nothing in the text cost the author anything (time, money, embarrassment, a Tuesday). Generated text is expense-free.
    • Hollow specificity: numbers that decorate rather than constrain ("boost productivity by 40%" with no denominator, no source, no consequence).
    • Symmetric enthusiasm: every clause at the same emotional temperature; no item loved more than another. Real people play favorites.
    • Both-sidesing: balanced hedged evenhandedness where a person would just have a position.
    • Tutorial cadence: explaining to the reader what the reader was promised, signposting, wrap-up endings, the shape of a lesson where the shape of a remark belongs.
    • Timeless floating: no now — no weekday, season, "this morning", nothing anchoring the text to a life in progress.
    • Perfect memory: recalls its own earlier points too neatly; humans drift, repeat one word too often, abandon a setup.
    • Category error against platform: reads like a blog post wearing a tweet, a press release wearing a Reddit post (when platform is given).
    • Borrowed expertise: fluent, correct command of a domain the rest of the text places out of the author's reach — the register, references, and life on display say novice, but the jargon lands like a specialist's. A ceramicist who suddenly discusses lock-free concurrency correctly. Real people signal the edge of their knowledge (they hedge, analogize, or defer); a paste-in of expertise does not.
  • Human counter-signals (credit when present, but remember they're gameable — Jakesch 2023: first person and contractions are exactly what fakes add): unfabricatable specifics, unresolved ambivalence, era-bound references, self-interruption that costs rhetorical polish, a detail that serves no persuasive purpose.
  • When torn, say torn. generated-leaning with reasons beats false confidence in either direction.

Read the full file on GitHub · 55 lines

Changes

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.

  1. 8d ago First seen · 55 lines · 64 tokens per session scan A c96a350b1c61

Subscribe to this mod's changes

idiolect-auditor is an agent published in the GitHub repository nagisanzenin/idiolect (23 stars, last pushed 2mo ago), licensed MIT. It adds 64 tokens to every session and 1,107 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 2026-09-01.