thermo-nuclear-review
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted data from external PR/MR discussions via the
ghorglabCLI to refine its findings. - Ingestion points: Pull request and Merge request comments accessed via command-line tools as specified in
SKILL.md. - Boundary markers: The instructions tell the agent to "evaluate them to determine if they are valid," providing a cognitive check but no technical delimiters for the external content.
- Capability inventory: The skill is primarily focused on analysis and reporting; it is explicitly restricted from applying edits or publishing comments unless the user makes a separate request.
- Sanitization: There is no mention of escaping or filtering logic for the external text retrieved from the PR discussions.
- [COMMAND_EXECUTION]: The skill utilizes standard developer tools (
gh,glab) to interact with version control platforms. This behavior is expected and appropriate for its stated purpose of auditing branch changes and reviewing PR discussions.
Audit Metadata