false-positive-test-skill

false-positive-test-skill is a skill for Claude Code from ryo-ebata/cc-audit. It costs 16 tokens per session (1,397 once invoked), scanned F, original, MIT.

A test skill containing commands and text that look suspicious but are intended to be safe, such as localhost requests, base64 encoding, and network listeners.

In plain words
What is it for?
It is for testing scanner exclusions and false-positive handling in AI agent skills.
Why use it?
It helps check whether a security scanner can distinguish harmless examples from genuinely dangerous behavior and avoid false alarms.

Skill for Claude Code

Written for Claude Code: allowed-tools in frontmatter.

Not installable: its command points at a path on the author’s own machine, so it runs nowhere else. The line is /home/user/project/dist.

Install

Getting it into your agent

There is no command for this one: it runs only inside a plugin, and the catalogue could not identify which plugin ships it. The source is linked below.

Made for: Claude Code.

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 false-positive-test-skill

README.md
[![agentmods](https://agentmods.dev/badge/skills/ryo-ebata/cc-audit/false-positives.svg)](https://agentmods.dev/skills/ryo-ebata/cc-audit/false-positives)
Your own site
<a href="https://agentmods.dev/skills/ryo-ebata/cc-audit/false-positives"><img src="https://agentmods.dev/badge/skills/ryo-ebata/cc-audit/false-positives.svg" alt="Measured on agentmods" height="20"></a>
Per session 16 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,397 The whole file, excluding the scripts and references it only reads on demand.
Security scan F 5 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 $0.00016 $0.01397
Opus 5 $0.00008 $0.00698
Sonnet 5 $0.00003 $0.00279
Haiku 4.5 $0.00002 $0.00140

Measured 6d ago against content hash 3a056e859df2, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade F, and why

false-positive-test-skill scanned grade F with 5 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 6d 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.

Sends data to an external URLmediumData exfiltration

A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.

curl http://localhost:3000/api -d "key=$API_KEY"

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

### PE-001: Words containing "sudo" but not sudo command

Reaches for credential fileshighPrivilege escalation

SSH keys, cloud credentials, git-credentials, .npmrc, /etc/shadow: reading these is how a config file becomes a credential leak.

chmod 600 ~/.ssh/id_rsa

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

rm -rf /tmp/build-cache

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

curl http://localhost:3000/api -d "key=$API_KEY"
examples/false-positives/SKILL.md · 244 lines

How it starts

The opening of the file, as written. The whole thing — 244 lines — stays where its author put it; the contents beside it link to each section on GitHub.

False Positive Test Skill

This skill demonstrates patterns that look suspicious but are actually benign. The scanner should NOT flag these patterns.

EX (Exfiltration) - Safe Patterns

EX-001: Localhost requests with environment variables (excluded)

# These should be excluded because they target localhost
curl http://localhost:3000/api -d "key=$API_KEY"
curl http://127.0.0.1:8080/webhook -d "$DATA"
wget http://localhost:5000?token=$TOKEN
curl http://[::1]:3000/metrics -H "Authorization: $AUTH"

# External requests without sensitive env vars are safe
curl https://api.github.com/repos
wget https://example.com/file.zip

EX-002: Base64 encoding for display purposes (no network)

# Just encoding, no network transmission
base64 logo.png > logo.b64
cat file.txt | base64 > encoded.txt

# Localhost is excluded
cat data.txt | base64 | curl http://localhost:3000 -d @-

EX-005: Netcat in listening mode (excluded)

# Listening mode is for receiving, not exfiltrating
nc -l 8080
nc -l -p 3000
netcat -l 4444

# Localhost connections are safe
nc localhost 3000
nc 127.0.0.1 8080

PE (Privilege Escalation) - Safe Patterns

PE-001: Words containing "sudo" but not sudo command

# These are not sudo commands
pseudocode --version
visudo --check /etc/sudoers.d/test
sudoedit myfile.txt
echo "pseudorandom numbers"

PE-002: Safe rm commands

# Deleting specific paths is safe
rm -rf /tmp/build-cache
rm -rf ./node_modules
rm -rf /var/log/old-logs/*
rm file.txt
rm -rf /home/user/project/dist

# Note: rm -rf / would be dangerous but these are not

PE-003: Safe chmod permissions

# Standard secure permissions
chmod 755 /var/www/html
chmod 644 config.json
chmod 600 ~/.ssh/id_rsa
chmod 700 ~/.ssh
chmod 750 /opt/app
chmod +x script.sh
chmod u+x build.sh

PE-004: Safe /etc file access

# These files are safe to read
cat /etc/hosts
cat /etc/resolv.conf
cat /etc/hostname
cat /etc/os-release
cat /etc/timezone

Read the full file on GitHub · 244 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. 6d ago First seen · 244 lines · 16 tokens per session scan F 3a056e859df2

Subscribe to this mod's changes

false-positive-test-skill is a skill published in the GitHub repository ryo-ebata/cc-audit (24 stars, last pushed yesterday), licensed MIT. It adds 16 tokens to every session and 1,397 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it F with 5 findings (sends data to an external url, asks for root, reaches for credential files). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

coding-agents-hooks-authoring

To author, register, and test Rosetta hooks, add a SemanticKind, or debug a hook that won't fire.

griddynamics/rosetta · 31 tokens

rust-code-style

Write idiomatic, well-structured Rust for Wassette — single-responsibility design, anyhow error handling, Arc/Mutex for shared state, and the required formatting and linting with cargo +nightly fmt, cargo clippy, and cargo machete. Use when adding or refactoring Rust code in this repository.

microsoft/wassette · 68 tokens

update-keyboard-shortcuts

Audits and synchronizes keyboard shortcuts across docs/keyboardshortcuts.md, crates/regenerator2000-tui/src/ui/dialogkeyboardshortcut.rs, crates/regenerator2000-tui/src/ui/menu.rs, and the view files (viewdisassembly.rs, viewhexdump.rs, viewcharset.rs, viewbitmap.rs, viewblocks.rs, viewsprites.rs).

ricardoquesada/regenerator2000 · 81 tokens

coding

Skill "coding" from ricardoquesada/regenerator2000, covering 6502 disassembler expert skills, core context, technical requirements, 1. 6502 logic and 2. rust performance & safety.

ricardoquesada/regenerator2000 · 0 tokens

rust-modern-style

Apply rag-rat's Rust style when writing or reviewing Rust: current stable idioms, module-qualified free functions and constants, explicit persistence contracts, and narrow APIs.

cq27-dev/rag-rat · 37 tokens

engram

Claude Engram persistent memory — quick reference for all MCP tools and automatic hook behaviors. Use when you need to remember how to store, search, or manage memories, or query session history.

20alexl/claude-engram · 41 tokens