Release v1.3.0 — CLI tools, CI/CD suite, 11 languages, security hardening · rahuldk1105/roast-my-codebase

What's new in v1.3.0

--explain [category] — Offline knowledge base: 38 categories with why-it-matters, how-to-fix, and code examples
--file <path> — Per-file drill-down: scan one file and show only its findings (great for pre-commit)
--breakdown — Score breakdown by category: see exactly which categories are costing the most points
--perf — Scanner timing: bar chart showing how long each scanner group took
--output-dir <path> — Save all 5 report formats (JSON/HTML/SARIF/JUnit/Markdown) in one command
--since-commit <sha> — Diff mode: only show findings for files changed since a git ref
--exit-codes — Meaningful exit codes: 2=threshold, 3=regression, 4=trend, 5=security, 6=license
--notify <url> — Slack/Teams/Discord webhook notifications with platform-native formatting
--update-readme — Auto-update health badge in README.md after every scan
--fail-on-trend — Smarter CI gating: only fail when score drops 3+ consecutive runs (with sparkline)
--bundle — Bundle size regression detection vs last recorded baseline
--monorepo — Scan each workspace package separately with per-package scores and rollup
--init — Interactive wizard to generate .roastrc.json
--serve — Live web dashboard with SSE auto-refresh on file changes

License compliance — Flag GPL/AGPL/LGPL deps in commercial projects
Database queries — Prisma/TypeORM/Sequelize/Mongoose anti-patterns (N+1, raw SQL, force:true)
Config auditor — tsconfig.json strictness, ESLint coverage, Prettier configuration
Test quality — Empty blocks, .only, always-true assertions, skipped tests, hardcoded delays
Bundle size — Track dist/ output size over time, alert on regressions
Custom Rule DSL — Define lint rules in .roastrc.json without writing TypeScript

Ruby, PHP, Swift, Kotlin — each with 3 scanners

Step Summary — Auto-writes markdown report to the Actions run page (zero config)
SARIF output — Findings appear as native Code Scanning alerts

100+ new roast entries, expanded to 8-12 per category
Combo roast when 4+ critical categories fire simultaneously
Contextual opening line based on score + finding count
Severity-aware large-file roasts
generateContextualVerdict — smarter verdicts for security/critical/test issues