pr-reference
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection Surface: This skill retrieves commit messages and unified diffs from the local repository and embeds them into an XML report. If a repository contains adversarial instructions within commit messages or code comments, these could be interpreted by the downstream AI agent. The skill uses XML tags and CDATA sections to help the agent distinguish between metadata and repository content.
- Command Execution through Git Tooling: The skill relies on local git installations to perform its core functions by invoking various git commands with user-provided parameters. While the scripts pass parameters as arguments to the git binary rather than executing them as raw shell strings, they interact with the local file system and repository state. The scripts include verification steps to ensure that git references exist before proceeding with data collection.
Audit Metadata