regression-guardlisted

# Regression Guard — Scan, Fix, Prevent When you fix a bug or quality issue, don't just fix it — add automated prevention so it can never happen again. This skill follows a systematic approach: scan the entire codebase for a class of problems, classify findings, batch fix, then add tooling to prevent regression. **Inspired by**: vm0 e7h4n's pattern — "修了就防回歸" Your args are: `$ARGUMENTS` Parse the args: - `scan <category>` — Scan codebase for a specific problem category - `guard <description>` — After fixing something, add regression prevention - `audit` — Full audit across all categories Categories: `type-safety`, `error-handling`, `security`, `testing`, `dead-code`, `performance`, `dependencies` --- ## Operation 1: Scan Systematically scan the codebase for a class of problems, classify by severity, produce an actionable report. ### Step 1: Choose Scanning Tools | Category | Tools | |----------|-------| | type-safety | `bun run lint`, `grep "as any\|@ts-ignore"` | | error-handling | `grep "try.*catch"`, `grep "console.error"` | | security | `grep "eval\|exec\|innerHTML"`, `grep "password\|secret\|token"` in non-.env | | testing | `grep "\.skip\|\.only"`, `grep "setTimeout" in *.test.ts` | | dead-code | `bun run knip`, `grep "// TODO\|// FIXME\|// HACK"` | | performance | `grep "JSON.parse.*JSON.stringify"` (deep clone), `grep "\.forEach"` in hot paths | | dependencies | `npm audit`, `bun run knip` | ### Step 2: Run Scans Execute ALL relevant scans and collect raw