zero-harm-ai-gha

Version updated for https://github.com/Zero-Harm-AI-LLC/zero-harm-ai-gha to version v1.0.7.

  • 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

The GitHub Action zero-harm-ai-gha detects AI-specific security risks in pull requests, such as prompt injection risk, secrets exposure, and unsafe LLM tool usage. It provides automated security checks to help developers prevent misuse of large language models (LLMs) by scanning code diffs before merge and emitting annotations for review. The action supports various output formats, including GitHub Actions annotations, JSON, Markdown, and SARIF.

What’s Changed

zero-harm-ai-gha v1.0.7

This release updates the GitHub Action branding and repository references to the new zero-harm-ai-gha name, and adds optional AI-mode support for zero-harm-ai-detectors while preserving the default heuristic behavior.

What’s new

  • renamed the GitHub Action display name to zero-harm-ai-gha
  • updated README, workflow examples, sample repo references, and Marketplace-facing copy to match the new name
  • updated sample workflow naming and findings artifact/output filenames
  • added ai-mode input to the GitHub Action
  • added --ai-mode flag to the CLI
  • PromptShield’s detector integration now enables AI mode by constructing AIConfig() for zero-harm-ai-detectors when requested
  • default behavior remains unchanged: if ai-mode is not enabled, detection continues to use heuristic mode

Usage

GitHub Action:

- uses: Zero-Harm-AI-LLC/zero-harm-ai-gha@v1
  with:
    ai-mode: true