pr-description
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from code changes which can lead to indirect prompt injection if the code being reviewed contains adversarial instructions.
- Ingestion points: The skill ingests data from external sources by executing
git logandgit diffto analyze the changes between branches. - Boundary markers: The prompt instructions do not include clear delimiters or specific instructions to the agent to treat the content of the code diffs as untrusted data rather than instructions.
- Capability inventory: The agent has the capability to execute shell commands and interact with the GitHub API via
gh pr create,gh pr edit, andgh pr readyto persist content to a remote repository. - Sanitization: No sanitization or filtering is applied to the output of the git commands before the data is used to generate the final PR description.
Audit Metadata