Private
Argus
Argus audits a codebase the way a careful senior reviewer would: cheap deterministic checks first, expensive AI reasoning only where the risk actually warrants it.
TypeScriptNext.jspgvectortree-sitter
Risk-ranked, not blanket
A local-first analysis funnel runs deterministic tree-sitter AST detectors plus ESLint, type-checking, and CVE scans across the repo, then a risk-ranked router decides which changed or risky code is worth sending to the AI review tier. That's what lets Argus audit codebases with thousands of files without the cost scaling with repo size.
A catalog that's actually caught things
A 50+ pattern detection catalog spans security, resilience, observability, deployment, and the specific failure modes of AI-generated code. A high-precision AST detector in that catalog caught a live API-key leak while the tool was being used on its own codebase.
A report you can act on
Findings roll up into a three-state readiness verdict — fix-before-launch, accepted-risk, or pass — instead of an undifferentiated wall of warnings, so a team can tell at a glance whether a change is actually safe to ship.
Interested in working together?
laksh@lakshjain.com