ai-detect

ai-detect is a skill for Claude Code from ne11nn/cantos-plugin. It costs 152 tokens per session (1,504 once invoked), scanned A, original, MIT.

An optional workflow that checks a text file with GPTZero, a third-party service that estimates whether writing may have been produced by AI. It also includes local analysis when the external scan is declined.

In plain words
What is it for?
Use it to scan a draft for AI-writing signals, analyze patterns locally, and update the file after review.
Why use it?
It makes the external upload explicit and provides a way to review the writing without sending it to that service.

Skill for Claude Code

Written for Claude Code: argument-hint in frontmatter. Also seen: reads .claude/ paths; mentions subagents.

Part of the cantos plugin — 25 skills, 2 commands, 11 agents, 2 hooks shipped together

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.

agentmods
npx agentmods add skills/ne11nn/cantos-plugin/ai-detect
Any agent
npx skills add ne11nn/cantos-plugin --skill ai-detect
Clone the repo
git clone --depth 1 https://github.com/ne11nn/cantos-plugin

Made for: Claude Code.

Or install cantos, the plugin that ships this one along with the rest of its 25 skills, 2 commands, 11 agents, 2 hooks.

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 ai-detect

README.md
[![agentmods](https://agentmods.dev/badge/skills/ne11nn/cantos-plugin/ai-detect.svg)](https://agentmods.dev/skills/ne11nn/cantos-plugin/ai-detect)
Your own site
<a href="https://agentmods.dev/skills/ne11nn/cantos-plugin/ai-detect"><img src="https://agentmods.dev/badge/skills/ne11nn/cantos-plugin/ai-detect.svg" alt="Measured on agentmods" height="20"></a>
Per session 152 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,504 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00152 $0.01504
Opus 5 $0.00076 $0.00752
Sonnet 5 $0.00030 $0.00301
Haiku 4.5 $0.00015 $0.00150

Measured 5d ago against content hash ca328f1394da, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

ai-detect 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 5d 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.

system/.claude/skills/ai-detect/SKILL.md · 135 lines

How it starts

The opening of the file, as written. The whole thing — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Skill: ai-detect

Invocation: /ai-detect <file-path>

Three-phase pipeline: browser scan (Haiku) → deep pattern analysis (Opus) → file updates (Sonnet).


This skill uploads the full draft text to a third-party website (gptzero.me) and requires the user's own GPTZero login in the browser window. The draft leaves the local machine. Before any upload:

  1. State plainly that running this skill sends the draft's text to GPTZero (gptzero.me), an external service, and needs the user to be logged in there.
  2. Ask the user to confirm they want to scan this specific draft externally. This is opt-in — never run it by default.
  3. Never scan sensitive, confidential, or private text (anything the user hasn't agreed to share externally — personal records, unpublished/embargoed work, client or third-party material) without an explicit go-ahead for that exact text. When in doubt, ask.
  4. If the user declines, do not open the browser or upload anything. Use the manual fallback below instead.

Manual fallback (when the user declines external scan)

Stay entirely local — no upload:

  1. Read the draft and apply references/signs-of-ai-writing.md by hand, flagging each passage that matches a catalogued pattern.
  2. Rewrite the flagged passages using the write-like-me skill (if installed) so the prose matches the user's own voice; otherwise rewrite to remove the catalogued tells while preserving citations, statistics, and word limits.
  3. Report which patterns were found and what was changed. Skip Phases 1–3 entirely — they exist only for the external GPTZero pipeline.

Phase 1 — Browser Scan (Haiku)

Only reach this phase after the Privacy and Consent gate above has passed (user explicitly agreed to upload this draft to GPTZero). If consent was not given, stop and run the manual fallback instead.

Spawn a Haiku agent for all browser work. Pass it the file path and content.

The Haiku agent should:

  1. Read the file at the given path — extract plain text (strip markdown formatting and headers for paste)
  2. Open a headed (non-headless) Playwright browser via the playwright-cli skill
  3. Navigate to gptzero.me. Take a snapshot to check login state
  4. If not logged in: output "Please log in to GPTZero in the browser window, then confirm here." Wait for user confirmation before continuing
  5. Navigate to the document scanner (Dashboard → Detect). Take a snapshot
  6. Find and click "Deep Analysis" or "Advanced Analysis" mode if visible. Take a snapshot to confirm
  7. Locate the text input area. Clear it. Paste the plain text content
  8. Click the Scan or Analyze button. Poll for results every 5 seconds (max 60 seconds total)
  9. Once results appear, extract from the DOM:
    • Overall AI probability score (%)
    • Each AI-flagged sentence + all reason tags + probability score
    • Human-rated sentences where AI probability is ≥ 10% only (skip any sentence rated ≥ 90% human — those are clean)
  10. Save raw results to /tmp/gptzero-raw-<timestamp>.md:

Read the full file on GitHub · 135 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. 5d ago First seen · 135 lines · 152 tokens per session scan A ca328f1394da

Subscribe to this mod's changes

ai-detect is a skill published in the GitHub repository ne11nn/cantos-plugin (1 stars, last pushed yesterday), licensed MIT. It adds 152 tokens to every session and 1,504 once invoked, about $0.0008 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.

Related

Other skills, from other repositories

stream-chain

Stream-JSON chaining for multi-agent pipelines, data transformation, and sequential workflows.

spencermarx/open-code-review · 18 tokens

run

Run the agents-concerto multi-agent development pipeline for a task — classify by complexity, implement in isolated worktrees (TDD + Tidy First), review under a two-party boundary, and stop at open PRs for human merge. Invoke explicitly with a task description; a weak or empty description is shaped interactively first…

moruno21/agents-concerto · 77 tokens

shape

Turn a rough idea into a well-formed task the pipeline can run — interrogate the ambiguity, derive testable acceptance criteria, flag contradictions, and write .agent-workspace/feature-request.md in the exact shape /run consumes. The upstream step before /agents-concerto:run. Invoke explicitly with a rough idea.

moruno21/agents-concerto · 68 tokens

setup

Bootstrap agents-concerto in the current project by interviewing the user for every configurable option — target repos (with per-repo PR host), tasksource, base branch, commit convention, and fix-cycle cap — then writing a filled .agent-workspace/config.md. Detects sensible defaults and asks for confirmation rather…

moruno21/agents-concerto · 73 tokens

foundry-hosted-agent-validation

Step-by-step process for validating a Python Foundry hosted agent sample (under python/samples/04-hosting/foundry-hosted-agents/) end to end — running it locally (native runtime and azd ai agent run) and after deploying it to an Azure AI Foundry project with azd. Use this when asked to validate a hosted agent sample.

microsoft/agent-framework · 82 tokens

python-feature-lifecycle

Guidance for package and feature lifecycle in the Agent Framework Python codebase, including stage meanings, feature-stage decorators, feature enums, and how to move APIs from one stage to the next.

microsoft/agent-framework · 43 tokens