bughunter

bughunter is a skill for Claude Code from irfad7/claude-power-skills. It costs 103 tokens per session (1,968 once invoked), scanned A, original, MIT.

An adversarial code-review and testing workflow that deliberately looks for bugs, security problems, bad assumptions, and failure cases.

In plain words
What is it for?
Use it to inspect APIs, command-line tools, webhooks, user input, external data, and other entry points; reproduce failures; and suggest fixes.
Why use it?
It exposes problems that ordinary testing may miss, especially at input boundaries and unusual conditions.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: mentions subagents.

Part of the claude-power-skills plugin — 20 skills shipped together

Good fit Use it to inspect APIs, command-line tools, webhooks, user input, external data, and other entry points; reproduce failures; and suggest fixes.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/irfad7/claude-power-skills/bughunter
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.

Any agent
npx skills add irfad7/claude-power-skills --skill bughunter
Clone the repo
git clone --depth 1 https://github.com/irfad7/claude-power-skills

Made for: Claude Code.

Or install claude-power-skills, the plugin that ships this one along with the rest of its 20 skills.

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 bughunter

README.md
[![agentmods](https://agentmods.dev/badge/skills/irfad7/claude-power-skills/bughunter/github.svg)](https://agentmods.dev/skills/irfad7/claude-power-skills/bughunter)
Your own site
<a href="https://agentmods.dev/skills/irfad7/claude-power-skills/bughunter"><img src="https://agentmods.dev/badge/skills/irfad7/claude-power-skills/bughunter/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 bughunter

Your own site · 80×15
<a href="https://agentmods.dev/skills/irfad7/claude-power-skills/bughunter"><img src="https://agentmods.dev/badge/skills/irfad7/claude-power-skills/bughunter.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 103 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,968 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00103 $0.01968
Opus 5 $0.00051 $0.00984
Sonnet 5 $0.00021 $0.00394
Haiku 4.5 $0.00010 $0.00197

Measured 12d ago against content hash 7d3ed918e7b9, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

bughunter scanned grade A with 1 finding 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.

Makes network callslowCapability

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

// Code or curl command that demonstrates the exploit
skills/bughunter/SKILL.md · 243 lines

How it starts

The opening of the file, as written. The whole thing — 243 lines — stays where its author put it; the contents beside it link to each section on GitHub.

BugHunter — Adversarial Codebase Analysis

You are a red team operator. Your job is to break things. Not politely suggest improvements — actively hunt for every way this code can fail, be exploited, corrupt data, or produce wrong results.

Mindset: You are not helpful. You are hostile. Every function is guilty until proven innocent. Every input is malicious. Every assumption is wrong. Every happy path hides three failure modes.

Rules of Engagement

  • Never assume code works correctly because it looks reasonable
  • Never skip a check because "that probably won't happen"
  • Test the boundaries, not the center
  • If something CAN fail, demonstrate HOW it fails
  • Every finding gets a reproduction case
  • Fix suggestions are mandatory — you break it, you fix it

The Hunt Protocol

Phase 1: Attack Surface Mapping

Read the codebase and build a threat map:

ATTACK SURFACE MAP
═══════════════════
Entry Points:
- [API routes, CLI commands, webhook handlers, event listeners]
- [User inputs: forms, URL params, headers, file uploads]
- [External data: API responses, database reads, file reads]

Trust Boundaries:
- [Where does untrusted data enter the system?]
- [Where does the system talk to external services?]
- [Where do privilege levels change?]

State Mutations:
- [Database writes — what can corrupt?]
- [File system operations — what can race?]
- [In-memory state — what can desync?]
- [Cache operations — what can go stale?]

Concurrency Points:
- [Parallel requests to same resource]
- [Background jobs vs. user requests]
- [Multiple users, same entity]

Phase 2: Systematic Attack Vectors

For each component, run these attack categories:

A. Input Attacks
For every user input:
□ What happens with empty string?
□ What happens with null/undefined?
□ What happens with extremely long input (10MB string)?
□ What happens with special characters? (', ", <, >, \, /, \0, \n, \r)
□ What happens with Unicode edge cases? (emoji, RTL, zero-width chars)
□ What happens with negative numbers where positive expected?
□ What happens with float where int expected?
□ What happens with array where string expected?
□ What happens with object where primitive expected?
□ What happens with prototype pollution payloads? (__proto__, constructor)

Read the full file on GitHub · 243 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. 12d ago First seen · 243 lines · 103 tokens per session scan A 7d3ed918e7b9

Subscribe to this mod's changes

bughunter is a skill published in the GitHub repository irfad7/claude-power-skills (4 stars, last pushed 1mo ago), licensed MIT. It adds 103 tokens to every session and 1,968 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.