greploop-apps
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of CLI tools (
gh,glab,git,p4) to interact with version control systems and external APIs. While these are used for their intended purpose, the instructions involve looping and automated decision-making based on external data. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (PR/MR/CL comments, descriptions, and review results) that could be controlled by an attacker.
- Ingestion points: Untrusted data enters the context via
gh pr view,glab mr view,gh api(issue comments), andglab api(discussions/notes) inSKILL.md. - Boundary markers: None identified. The skill directly parses and acts upon the text found in review comments.
- Capability inventory: The skill has the capability to execute shell commands (
Bash), write to the filesystem through automated fixes, and perform network operations viagit push,p4 shelve, and platform-specific API calls. - Sanitization: There is no explicit sanitization or filtering of the content retrieved from Greptile or other users' comments before the agent is instructed to 'Read the file and understand the comment in context' and 'make the fix'.
Audit Metadata