greploop
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill extensively uses the Bash tool to run shell commands for version control and platform interactions, including git, gh, glab, and p4. These commands are used to identify pull requests, push code updates, and poll for review status.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon external feedback from the Greptile code review tool, which presents a surface for indirect prompt injection. 1. Ingestion points: The skill reads pull request/merge request descriptions and comments via platform APIs and CLI tools (gh pr view, glab api, p4 describe) in SKILL.md and references/gitlab-api.md. 2. Boundary markers: No specific boundary markers or instructions to ignore malicious content within the ingested review comments are present. 3. Capability inventory: The agent has the capability to write to the local filesystem (to apply code fixes) and perform network operations via VCS push/shelve commands (git push, p4 shelve). 4. Sanitization: The instructions do not specify any sanitization or validation logic for the text retrieved from review comments before it is used to inform code changes.
Audit Metadata