# Vexa Security (vexasec.io) > Local-first security for software dependencies and AI-agent actions. Built by Noviq Technologies. ## Core Products (Adopt Independently) 1. **Vexa SCA** (Tier 1 GA — Node.js / JS / TS: `package-lock.json` v2/v3, `npm-shrinkwrap.json`, `pnpm-lock.yaml` v5/v6/v9; Tier 2 Experimental: `yarn.lock`, `requirements.txt`, `Cargo.lock`, `go.sum`, `pom.xml`): Evidence-aware Software Composition Analysis. Parses lockfiles into deterministic transitive dependency graphs, surfaces heuristic repository usage signals, marks unverified paths explicitly as `UNKNOWN` with reason codes instead of guessing, and previews same-major remediations (falling back to major bumps only when required). Cloud mode (`vexa-sca-cloud`) supports zero-clone GitHub App lockfile scans or sanitized CLI bundle uploads with zero source code retention. - Install: https://vexasec.io/sca/install - Docs: https://vexasec.io/sca/docs - Product page: https://vexasec.io/sca 2. **Vexa Arc** (Community): Local-first policy boundary for AI coding assistants (Claude Code, Cursor) and stdio MCP tools. Enforces host hooks and generates CycloneDX AI-BOMs. - Install: https://vexasec.io/arc/install - Docs: https://vexasec.io/arc/docs - Product page: https://vexasec.io/arc ## Documented Security & Network Boundaries - **Local-first by default:** CLI analysis runs locally. Network access is limited to explicit advisory lookups (Vexa SCA) and local policy evaluation (Vexa Arc). - **Zero source code retention:** Cloud SCA dashboard syncs sanitized dependency graphs and metadata, not raw source code. - **Product Truth Matrix:** https://vexasec.io/trust - **Security Boundaries:** https://vexasec.io/security ## Comparisons & Pricing - Pricing & Editions: https://vexasec.io/pricing - Vexa vs. Snyk: https://vexasec.io/vs/snyk - Vexa vs. Socket: https://vexasec.io/vs/socket - Vexa vs. Aikido: https://vexasec.io/vs/aikido