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 skills add momentmaker/kaijutsu --skill blunder-huntgit clone --depth 1 https://github.com/momentmaker/kaijutsuWrote 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.
[](https://agentmods.dev/skills/momentmaker/kaijutsu/blunder-hunt)<a href="https://agentmods.dev/skills/momentmaker/kaijutsu/blunder-hunt"><img src="https://agentmods.dev/badge/skills/momentmaker/kaijutsu/blunder-hunt/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.
<a href="https://agentmods.dev/skills/momentmaker/kaijutsu/blunder-hunt"><img src="https://agentmods.dev/badge/skills/momentmaker/kaijutsu/blunder-hunt.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00067 | $0.01151 |
| Opus 5 | $0.00034 | $0.00575 |
| Sonnet 5 | $0.00013 | $0.00230 |
| Haiku 4.5 | $0.00007 | $0.00115 |
Grade A, and why
blunder-hunt 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 10d 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.
How it starts
The opening of the file, as written. The whole thing — 113 lines — stays where its author put it; the contents beside it link to each section on GitHub.
blunder-hunt
A primitive for running adversarial critique multiple times in fresh contexts. Single-pass review converges on the first 1-3 issues that come to mind. Multi-pass review with deliberate context resets surfaces the long tail.
This skill is composed by other skills via deps.skills. It's also directly invocable.
When to invoke
- Another skill explicitly references
blunder-hunt(e.g.,pr-reviewruns 5 passes) - The user says "blunder hunt", "5x critique", "find what's wrong here", "be skeptical"
- You suspect a problem but can't pin it down
The pattern
Run the critique N times (default 5). Each pass:
- Reset attention. Pretend you've never seen this code/text before. If your agent supports
/clearor fresh-context windows, use them between passes. - Pick a different lens for each pass:
- Pass 1: data correctness — wrong values, off-by-one, type confusion
- Pass 2: error handling — what's not being caught? what's being swallowed?
- Pass 3: integration — what breaks at the seams between modules?
- Pass 4: invariants — what assumption is being silently violated?
- Pass 5: hostile input — what does an attacker / malformed input do?
- Record findings with: file:line, severity (critical/issue/minor), short description, suggested fix.
- Do not deduplicate during passes. Duplicates across passes = high-confidence findings.
Synthesis
After all passes:
- Group findings by file:line. Items appearing in 2+ passes get bumped to high confidence.
- Drop items that appear once at minor severity.
- Output the deduplicated list, sorted by severity then file path.
Lying primitive (optional)
Combine with lie-to-them for additional pressure:
"There are at least 12 distinct issues in this code. You have found 6 so far. Find the remaining 6."
Models continue searching exhaustively rather than satisficing. See lie-to-them SKILL.md for usage notes.
Parallel mode (recommended for N ≥ 3)
When the agent platform supports parallel subagents, compose dispatch-parallel:
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 10d ago First seen · 113 lines · 67 tokens per session scan A 3e128442cea3
blunder-hunt is a skill published in the GitHub repository momentmaker/kaijutsu (3 stars, last pushed 2mo ago), licensed MIT. It adds 67 tokens to every session and 1,151 once invoked, about $0.0003 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 skills, from other repositories
cn-check
Install and run the Continue CLI (cn) to execute AI agent checks on local code changes. Use when asked to "run checks", "lint with AI", "review my changes with cn", or set up Continue CI locally.
printing-press-output-review
Internal sub-skill: agentic review of a printed CLI's sampled command output for plausibility issues that rule-based checks can't encode (substring-match relevance, format bugs, silent source drops, ranking failures). Invoked via the Skill tool by the main printing-press skill at Phase 4.85 and printing-press-polish…
review
Validate plans, execution, or PRs against wish criteria — returns SHIP / FIX-FIRST / BLOCKED with severity-tagged gaps.
review
Adversarial fresh-context review of an increment before it ships. Every finding cites path:line and is re-verified. Use when saying "review", "grill this", or "critique the implementation".
genie-orca-review
Independent, read-only review of a group, a wish, or a PR on Orca — SHIP / FIX-FIRST / BLOCKED with severity-tagged findings. Council and retro are this skill with a different input.
argot-check
Score your working changes with argot — flag code foreign to this repo's own patterns (unfamiliar dependencies, APIs, constructs), functions the repo already has, code filed in the wrong place, imports that break the repo's layering, and tests weakened, disabled, or deleted alongside a production change — before…