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 escoffier-labs/brigade --skill bug-huntgit clone --depth 1 https://github.com/escoffier-labs/brigadeWrote 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/escoffier-labs/brigade/bug-hunt)<a href="https://agentmods.dev/skills/escoffier-labs/brigade/bug-hunt"><img src="https://agentmods.dev/badge/skills/escoffier-labs/brigade/bug-hunt.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Memory Poisoning · line 19 Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.Fix: Protect agent memory and state from modification by untrusted content. Use read-only memory for critical instructions and validate all state changes.
- medium Excessive Agency · line 71 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00045 | $0.00897 |
| Opus 5 | $0.00023 | $0.00449 |
| Sonnet 5 | $0.00009 | $0.00179 |
| Haiku 4.5 | $0.00005 | $0.00090 |
Grade A, and why
bug-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 8d 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- bug-hunt — 94% identical, 2 lines differ
How it starts
The opening of the file, as written. The whole thing — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
bug-hunt
A correctness sweep that only reports bugs it failed to refute. Finders generate candidates; verifiers try to kill them; survivors make the report. The single biggest failure mode of agent bug-hunting is plausible-but-wrong findings, so verification is not optional.
Read-only. Finding bugs and fixing them are separate engagements.
Lenses
Sweep with each lens. With parallel subagents available, one finder per lens; otherwise sequential passes.
| Lens | Hunting for |
|---|---|
| Logic | Inverted conditions, off-by-one, wrong operator, unreachable branches, broken invariants |
| Error handling | Swallowed exceptions, missing error paths, errors that corrupt state before propagating, misleading messages |
| Edge cases | Empty/nil/zero inputs, unicode, huge inputs, boundary values, first/last iteration |
| Concurrency | Races, missing locks, shared mutable state, TOCTOU, async ordering assumptions |
| API misuse | Contract violations against libraries and the project's own interfaces, ignored return values, resource leaks, lifecycle errors |
Focus finders on code that is reachable and load-bearing: entry points, hot paths, recently changed files (git log --since is a good prior). A bug in dead code is info, not a finding.
Two deep lenses live as their own skills: latent-premises for unguarded assumptions that hold today with no trigger yet, and retry-safety for side effects that double-apply on a re-run. Reach for them when the sweep keeps surfacing "could break later" candidates that fail the trigger rule.
Verification (mandatory)
Every candidate gets an adversarial pass before it may appear in the report. The verifier's job is to REFUTE the finding, default skeptical:
- Read the actual code path end to end, including callers and guards the finder may have missed.
- Trace a concrete input that triggers the bug. No trigger, no bug.
- Check whether a test, type system, or runtime check already prevents it.
- Verdict: confirmed (with the triggering scenario), refuted (drop silently), or unverifiable (report downgraded one severity, marked
(unverified)).
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.
- 8d ago First seen · 75 lines · 45 tokens per session scan A c023e8d6fb93
bug-hunt is a skill published in the GitHub repository escoffier-labs/brigade (72 stars, last pushed today), licensed MIT. It adds 45 tokens to every session and 897 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 2026-08-30.
Other skills, from other repositories
engraphis-memory
Give the agent durable, scoped, explainable memory across sessions and repositories through the Engraphis MCP tools. Use when you learn a convention, decision, bug cause/fix, or user preference worth keeping; when prior context would help before you answer or act (to avoid re-asking or re-deriving); when asked "why is…
github-issue-solve
Solve a GitHub issue by collecting context, implementing a fix, and opening or updating a pull request.
memorywhale
Query and write durable debugging memory recorded by MemoryWhale. Use when debugging a failure that may have happened before, when you need the exact error/flags/output from an earlier attempt, when the user asks "how did we fix this last time?", or once you've figured out why something failed / how a fix worked and…
Sverklo Code Intelligence
Gives you deep codebase understanding — semantic search, blast-radius analysis, PR review, and health audits.
javascript-sandbox
Best practices for using the clodex built-in JavaScript sandbox for browser debugging, fetched-data processing, attachments, and mini-app orchestration within its bundled capability boundary.
graphify
Use to orient in an unfamiliar codebase with graphify, a local tree-sitter code graph — "what does this symbol touch", "how do these two connect", "trace this call chain". Opt-in (the human installs graphify once). A forward-orientation aid, NOT a reliable reverse-dependency tool. Shipped and kept current by Loom.