# Bug Hunt

> Adversarial multi-agent bug hunting pipeline. Use when the user says "hunt for bugs", "adversarial bug review", "bug hunt pipeline", "find and verify bugs", or wants a thorough multi-phase bug analysis with verification. Runs three context-isolated agents (Hunter, Skeptic, Arbiter) to find bugs, challenge them, and deliver verified results.

- Skill: `tjboudreaux/bug-hunt` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add tjboudreaux/bug-hunt`
- Raw SKILL.md: https://api.skillmd.com/api/skills/tjboudreaux/bug-hunt/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: tjboudreaux (https://skillmd.com/u/tjboudreaux)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/tjboudreaux/bug-hunt

---


# Adversarial Bug Hunt Pipeline

Run a 3-phase adversarial bug hunting pipeline using context-isolated subagents. Each phase runs in a fresh context to prevent bias contamination.

## Phase Overview

1. **Bug Hunter** — Aggressively scans codebase for all potential bugs, scored by severity
2. **Bug Skeptic** — Attempts to disprove each reported bug with counter-evidence
3. **Bug Arbiter** — Weighs both arguments and delivers final verdicts

## Determine Scope

Before launching Phase 1, determine the target scope:

- If the user specified files or directories, use those exact paths
- If invoked in a project context with no specific scope, scan the full codebase
- Use Glob to verify the target paths exist

Store the resolved scope as a comma-separated list of paths for passing to subagents.

## Phase 1: Bug Hunter

Read the hunter prompt from `references/hunter-prompt.md` in this skill's directory.

Spawn a subagent with these parameters:
- **Type:** `feature-dev:code-reviewer`
- **Prompt:** Include the full hunter prompt content, followed by the target scope:

```
<hunter-prompt content here>

## Target Scope

Analyze the following files/directories: <resolved paths>
```

Store the Hunter's complete output as `HUNTER_REPORT`.

Display a brief status update: "Phase 1 complete — Hunter found X bugs. Starting adversarial review..."

## Phase 2: Bug Skeptic

Read the skeptic prompt from `references/skeptic-prompt.md` in this skill's directory.

Spawn a subagent with these parameters:
- **Type:** `feature-dev:code-reviewer`
- **Prompt:** Include the full skeptic prompt content, followed by the Hunter's report:

```
<skeptic-prompt content here>

## Bug Report to Review

The following bugs were reported by the Bug Hunter agent. Review each one:

<HUNTER_REPORT>
```

Store the Skeptic's complete output as `SKEPTIC_REPORT`.

Display a brief status update: "Phase 2 complete — Skeptic reviewed all bugs. Starting final arbitration..."

## Phase 3: Bug Arbiter

Read the arbiter prompt from `references/arbiter-prompt.md` in this skill's directory.

Spawn a subagent with these parameters:
- **Type:** `general-purpose`
- **Prompt:** Include the full arbiter prompt content, followed by both reports:

```
<arbiter-prompt content here>

## Hunter's Bug Report

<HUNTER_REPORT>

## Skeptic's Review

<SKEPTIC_REPORT>
```

Store the Arbiter's complete output as `ARBITER_REPORT`.

## Present Final Report

Display the Arbiter's full output to the user. This is the final consolidated report containing:
- Each bug with verdict (REAL BUG or NOT A BUG)
- Confidence levels
- Summary statistics
- Ranked list of confirmed bugs by severity

Preface the report with a brief summary line:
"Adversarial bug hunt complete. X bugs confirmed out of Y reported."