Agent Guard Secret Guardrails
Version updated for https://github.com/JeongJaeSoon/agent-guard to version v2.0.1.
- This action is used across all versions by ? repositories.
Action Type
This is a Composite action.
Go to the GitHub Marketplace to find the latest changes.
Action Summary
Agent Guard is a deterministic guardrail designed to prevent AI coding agents from accidentally exposing sensitive information, such as .env files containing private keys or credentials. It monitors the agent’s interactions with these files at runtime and blocks them if they attempt to read or write sensitive data before it can be used. This tool helps in real-time detection of potential leaks by using gitleaks for scanning and plain shell scripts for integration, ensuring defense-in-depth security measures are implemented.
What’s Changed
What’s Changed
- fix(hooks): scan NotebookEdit cell content for secrets and PII by @JeongJaeSoon in https://github.com/JeongJaeSoon/agent-guard/pull/113
- chore(release): v2.0.1 by @JeongJaeSoon in https://github.com/JeongJaeSoon/agent-guard/pull/114
Full Changelog: https://github.com/JeongJaeSoon/agent-guard/compare/v2.0.0...v2.0.1