Challenges code changes, design documents, and review reports to find missed risks.
npx skills add https://github.com/HACK-WU/skills --skill challenger-hack-wu
Please help me install this Agent Skill. Skill: challenger Source: https://github.com/HACK-WU/skills/tree/main/skills/challenger Command: npx skills add https://github.com/HACK-WU/skills --skill challenger-hack-wu
High-intent search queries and answers about installing and using this skill.
Run an adversarial challenge pass that assumes the approval is wrong and hunts for evidence. Verify the root cause was fixed, check callers handle new behavior, search for similar unfixed patterns, and grade each finding by risk level.
Adversarial design review assumes an approved design still has gaps and challenges goal alignment, feasibility, scenario completeness, and risk coverage. It outputs objections with reasoning chains, alternatives, and acknowledged strengths rather than a simple issue list.
Apply data conservation checks: trace where modified data flows, verify quantities do not silently shrink through filters or swallowed exceptions, and confirm field content survives serialization. Every exception handler must show where the dropped data actually went.
Yes, it accepts analysis or review reports from other skills as the challenge target. It focuses on caller-specified points such as whether the root cause is correct, the impact scope is complete, and the risk assessment is over- or under-stated.
Single reviews often miss piggyback changes, cross-module impacts, and extreme-condition failures. Use a second-pass challenge when changes touch critical functionality, data pipelines, or when the first review reported no blocking issues.
iclinicFinalize features by merging, creating a PR, or summarizing completion.
apacheIdentify and merge duplicate issues in issue trackers with maintainer confirmation.
hg-pyunGenerate consensus-approved implementation plans through architect and critic panel evaluation.
OutlineDrivenAutomates formal verification plans for software features with property-based testing.
