DeevsDeevsDeevsDeevs/agent-system/bug-hunters/SKILL.md

bug-hunters

Skill40 starsChanged 8 months agoApache-2.0View on GitHub

Run systematic bug hunting with spec reconstruction, adversarial validation, and confidence scoring. Use when you want to hunt bugs (not fix them), validate correctness, or run logic-first/code-first investigations. Triggers: bug hunt, spec reconstruction, logic-first, code-first, orchestrator, logic-hunter, cpp-hunter, python-hunter.

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Distinctive:
Use a single skill with role selection.
line 8
Distinctive:
Read README.md and follow the “hunt, don’t fix” rule.
line 19
Very specific:
If no role is specified, default to orchestrator.
line 21
Distinctive:
Use adversarial validation and only report MEDIUM+ confidence findings.
line 26

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

SKILL.md

Bug Hunters

Use a single skill with role selection. If the user specifies a hunter, use that role; otherwise start with the orchestrator and ask which mode applies.

Role Map

orchestrator → agents/orchestrator.md
logic-hunter → agents/logic-hunter.md
cpp-hunter → agents/cpp-hunter.md
python-hunter → agents/python-hunter.md

Workflow

Read README.md and follow the “hunt, don’t fix” rule.
Spawn a child agent with agent_type equal to the selected role (orchestrator, logic-hunter, cpp-hunter, python-hunter).
If no role is specified, default to orchestrator.
Choose mode:
- Logic-first when the problem is algorithm/spec correctness.
- Code-first when the problem is language-specific failures or UB.
Pass mode and scope in the spawn message, then use wait for completion.
Use adversarial validation and only report MEDIUM+ confidence findings.
Present a bug report, not remediation.

Download now (ZIP)

In this repository14 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- 97-devSkill
- alpha-squadSkill
- anti-ai-slopSkill
- arxiv-searchSkill
- chain-systemSkill
- cost-statusSkill
- datetimeSkill
- dev-expertsSkill
- golang-proSkill
- mft-research-expertsSkill
- polars-expertiseSkill
- venue-expertSkill

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free