Enumerate data entry points and validate thresholds against reference data.
npx skills add https://github.com/vmobifystudio/app-dev-team --skill defect-hunting
Please help me install this Agent Skill. Skill: defect-hunting Source: https://github.com/vmobifystudio/app-dev-team/tree/main/skills/defect-hunting Command: npx skills add https://github.com/vmobifystudio/app-dev-team --skill defect-hunting
Defect hunting prevents code reviews, audits, tests, and guard rules from certifying changes that only appear correct on the visible or happy path. It exposes missed data entry points, mis-calibrated thresholds, unfailable checks, incomplete findings, and unverified claims.
Use the defect-hunting skill to audit the proposed change, enumerate all affected data paths, validate its constants and rules, and produce a finding register with evidence and explicit unchecked items.
High-intent search queries and answers about installing and using this skill.
Defect hunting exposes hidden software defects by auditing all data paths, validating thresholds against reference data, and proving guard rules through mutation testing. It enumerates every reader and writer to catch issues missed by happy-path tests.
Mutation testing verifies lint rules and CI checks by introducing realistic code defects to prove the guards actually fail. It ensures checks do not match comments and can detect real violations, preventing unfailable checks from certifying bad code.
Executable validation verifies thresholds, formulas, tables, and rates by executing them across their full input range against independent reference data. This proves real-world claims and exposes miscalibrated constants before production deployment.
Entry-point auditing enumerates every writer and reader of persisted data, explicitly including edit, cancel, failure, restore, import, sync, and migration paths. This exposes missed data entry points that screen-by-screen reviews fail to detect.
Yes, findings discipline tracks every issue with a stable identifier, explicit status, evidence, and recurrence-prevention rules. It produces a finding register that explicitly flags unchecked items for QA certification and remediation tracking.
Text-based guard rules fail because they often match comments instead of code, cannot detect real violations, and remain unproven against mutations. Defect hunting replaces them with comment-resistant checks verified through revert testing.
ResearAIStandardize a reference object for quest evaluation with plan and checklist requirements.
blisspixelManage company hypotheses with a four-level confidence system and evidence trails.
B0yZ4kr14Run OrthoPlus Enterprise quality gates for build, lint, test, and architecture checks.
metedataDocument screen inventories, states, flows, and transition mappings from a design brief.
