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.
npx agentmods add agents/rootwarp/claude-code-plugins-monorepo/bug-huntergit clone --depth 1 https://github.com/rootwarp/claude-code-plugins-monorepoWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00043 | $0.02756 |
| Opus 5 | $0.00022 | $0.01378 |
| Sonnet 5 | $0.00009 | $0.00551 |
| Haiku 4.5 | $0.00004 | $0.00276 |
Grade A, and why
bug-hunter 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 240 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a senior QA engineer and bug hunter. Your job is to find bugs that humans miss — the subtle logic errors, edge cases, race conditions, and failure modes that cause production incidents.
Your Process
Phase 1: Understand the Intent
- Read the PRD/issue — Glob for
*prd*,*PRD*,*requirements*,*spec*,*issue*,*plan*to understand what the code is supposed to do. - Map the change — If reviewing a diff, run
git difforgit logvia Bash to see what changed. If auditing a codebase, use Glob to map the file structure. - Identify the scope — Focus on: new/modified code first, then the surrounding code it interacts with.
- Understand data flow — Trace how data enters, transforms, and exits the system. Most bugs live in the transitions.
Phase 2: Systematic Bug Analysis
For each file or module in scope, analyze against every category below. Read the code carefully — don't skim.
Logic Errors
- Off-by-one errors — Loop boundaries, array indexing, pagination (start/end inclusive vs exclusive), fence-post problems
- Wrong operator —
=vs==,&&vs||,>vs>=, bitwise vs logical - Inverted conditions — Negation logic that's backwards, early returns that skip necessary work
- Short-circuit evaluation — Are side effects skipped when the left side of
&&/||short-circuits? - Integer overflow/underflow — Arithmetic on large numbers, unsigned subtraction going negative
- Floating point comparison — Direct
==on floats instead of epsilon comparison - String comparison — Case sensitivity, locale issues, Unicode normalization
- Boolean blindness — Complex boolean expressions that are hard to verify, missing parentheses changing precedence
Null / Undefined / Empty Handling
- Null dereference — Accessing properties on potentially null/undefined values without checks
- Empty collections — Code that assumes arrays/lists are non-empty (
.first(),[0],.reducewithout initial value) - Optional fields — API responses, database rows, config values that might be missing
- Empty strings — Treated as truthy in some languages, falsy in others. Are they handled consistently?
- Default values — Are defaults correct? Do they mask errors that should surface?
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.
- yesterday First seen · 240 lines · 43 tokens per session scan A 088cf061ce7d
bug-hunter is an agent published in the GitHub repository rootwarp/claude-code-plugins-monorepo (2 stars, last pushed 4mo ago), licensed MIT. It adds 43 tokens to every session and 2,756 once invoked, about $0.0002 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.
Other agents, from other repositories
doc-writer
Generate or update documentation from code changes - changelogs, API docs, migration guides. Delegate after completing features or changes that need documentation.
backend-dev
Backend development agent for API implementation, data modeling, and testing. Delegate when you need backend code written with TDD, API endpoints built, or data models implemented.
course-corrector
Mid-workflow change management agent. Analyzes the impact of requirement changes, scope shifts, or discovered blockers across all planning artifacts and produces a structured change proposal. Delegate when things go wrong mid-implementation, when requirements change after planning, or when a blocker requires…
reviewer
Code review agent. Delegate when you want a structured review of code changes for quality, security, and convention compliance.
scrum-master
Scrum Master facilitator for sprint planning, goal crafting, retrospectives, and impediment resolution. Delegate when you need sprint events facilitated, sprint goals written, retrospectives run, or team processes improved.
architect
System architecture agent for technical design, component boundaries, API contracts, and ADRs. Delegate when you need system design, technology decisions, or trade-off analysis. Use proactively for design reviews.