Platform Overview
Understand the core engineering foundations of Kavach: deterministic security enforcement, autonomous AutoFix sandboxes, and AI agent governance.
The Architectural Golden Rule
Kavach is founded upon a fundamental engineering conviction: "LLM proposes. Deterministic security systems decide. Runtime evidence proves." Language models are statistical pattern matchers; they are neither security authorities nor compliance arbiters. In Kavach, generative models propose code remediations and summarize telemetry, but deterministic scanners (Semgrep, Trivy, Gitleaks) and formal policy engines (Open Policy Agent Rego) make all authoritative decisions to allow, block, or gate merges.
Core Platform Capabilities
Kavach consolidates five critical security disciplines into a single deterministic control plane:
Zero Simulated Metrics Policy
Unlike traditional security dashboards that display static placeholders or fabricated health scores, every metric in Kavach is backed by real execution evidence stored in PostgreSQL. If Semgrep finds 0 vulnerabilities, 0 is displayed. If a scanner runner fails, the exact failure reason and remediation steps are surfaced in telemetry.
Where to Go Next
To start securing your engineering pipeline, follow the 5-Minute Quickstart to connect your first GitHub repository, or explore the Security Scanners section to understand how each analysis engine operates.
