Show HN: Offense – Security tools for your agents

Category: security

Tags: security, ai-agents, mcp, dependency-scanning, static-analysis

Score: 6.0/10 (Innovation: 5, Technical: 6, Documentation: 8, Utility: 5)

Offense provides security tooling (dependency auditing, source scanning, GitHub Actions inspection) that coding agents can discover and invoke via CLI, MCP, or HTTP, returning structured evidence for the agent to act on. It combines OSV advisory lookups with Opengrep and Zizmor scanners in a job-based workflow with local, self-hosted, and hosted modes. The project is notable for its agent-first design and explicit emphasis on scope, receipts, and verification.

Target audience: backend devs, security engineers, devops, AI agent developers

Repository: https://github.com/offense-research/offense · JavaScript · Apache-2.0

View on Hacker News