What problem does it solve? Reproducing a bug and proving it is a genuinely reportable vulnerability are different questions; this Skill puts every confirmed finding through an adversarial trial so only findings that survive cross-examination reach the final report. ## Core Features & Use Cases - Adversarial Trial Workflow: A prosecution subagent argues each finding is real and exploitable, a defense subagent argues the strongest case against it, and a judge subagent independently re-checks the evidence and rules guilty or not-guilty. - Idempotent Verdict Recording: Verdicts are written back into report.json and api_findings.json plus an append-only courtroom_verdicts.json transcript, so re-runs never re-try decided cases. - Standardized Report Generation: Guilty findings receive a fixed-format write-up at report/<package>/<finding-id>-<slug>.md with reproduction steps, impact, remediation, and MASVS/MASTG references. - Use Case: After dynamic verification and API testing confirm a dozen candidate findings in an authorized Android pentest, run the trial to filter out findings with unrealistic preconditions and produce defensible final reports for the survivors. ## Quick Start Run the courtroom verdict on the confirmed findings for target hash <sha256> and write up the guilty verdicts as final reports.
