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 chrisallenlane/claude-swe-workflows --skill lead-bug-huntgit clone --depth 1 https://github.com/chrisallenlane/claude-swe-workflowsWrote 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/chrisallenlane/claude-swe-workflows/lead-bug-hunt)<a href="https://agentmods.dev/skills/chrisallenlane/claude-swe-workflows/lead-bug-hunt"><img src="https://agentmods.dev/badge/skills/chrisallenlane/claude-swe-workflows/lead-bug-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/chrisallenlane/claude-swe-workflows/lead-bug-hunt"><img src="https://agentmods.dev/badge/skills/chrisallenlane/claude-swe-workflows/lead-bug-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 | $0.00104 | $0.06669 |
| Opus 5 | $0.00052 | $0.03334 |
| Sonnet 5 | $0.00021 | $0.01334 |
| Haiku 4.5 | $0.00010 | $0.00667 |
Grade A, and why
lead-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 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 — 475 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Lead-Bug-Hunt — Autonomous Bug Elimination
Drives a codebase toward "no bugs above severity floor" without operator involvement between startup and termination. The operator states scope, severity floor, constraints, and finisher preference at startup; the skill then loops /bug-hunt → triage → /implement-batch until two consecutive hunt passes produce no findings above the floor, or until a hard cap or andon cord halts the run.
This skill is a narrower sibling of /lead-project. /lead-project takes open-ended commander's intent and decides which skills to invoke from a broad repertoire. /lead-bug-hunt has a fixed loop shape (hunt → fix → re-hunt) and a bounded sub-skill repertoire. It exists because "iterate until bugs converge" is a recurring workflow worth canonizing.
Philosophy
This skill implements the autonomy discipline documented in references/autonomy.md. The shared discipline governs the five levers (altitude rule, pre-loaded options, pre-rebutted recommendation, commander's intent, risk budgets), the cascade rule, the no-unilateral-breaking-changes guardrail, and the shared handoff template. Skill-specific shape (the hunt-cycle structure, convergence criteria, severity-floor triage) is layered on top.
The loop converges on a floor, not on zero
Bugs below a stated severity floor do not block termination. They are recorded as deferred items in the completion report. The operator chooses the floor at startup; defaults to High (fix Critical and High; defer Medium and below).
This is the dominant judgment call this skill exposes. Setting the floor too low produces a loop that never converges (/bug-hunt always finds something at Low). Setting it too high produces a run that ships with real bugs unaddressed. The floor is elicited explicitly so the operator owns this trade-off.
Reproducing tests are durable artifacts
/bug-hunt commits reproducing tests as it works. Those tests outlive the run — they become permanent regression tests in the suite. Their quality matters more than typical test code because they will catch regressions for years and bias future test authors who read them as examples.
What ships with it
1 file 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 · 475 lines · 104 tokens per session scan A 45d2f2ed07ec
lead-bug-hunt is a skill published in the GitHub repository chrisallenlane/claude-swe-workflows (18 stars, last pushed 3mo ago), licensed MIT. It adds 104 tokens to every session and 6,669 once invoked, about $0.0005 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
afc:issue
Analyze GitHub issue — investigate bug reports, understand requirements, inspect issues.
afc:doctor
Diagnose afc plugin setup and health.
afc:debug
Bug diagnosis and fix — root-cause analysis for errors, crashes, broken behavior.
triage-issue
Analyze a GitHub issue, verify claims against the codebase, and close invalid issues with a technical response.
hive.error-recovery
Follow a structured recovery decision tree when tool calls fail instead of blindly retrying or giving up.
review-loop
Run the adversarial verification loop — implement, then hand the change to a fresh checker that did not write it, fix what it finds, and re-dispatch until APPROVE. Use before claiming any behavioural change is done, and on requests like "review loop", "adversarial review", "independent review", "get this verified"…