assess-pull-request
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes legitimate developer tools (
gh,glab,git) to interact with source control providers. It includes specific guidance for handling authentication constraints (e.g., using temporary HTTPS clones when SSH is restricted) and identifies standard commands for retrieving PR metadata and diffs. - [EXTERNAL_DOWNLOADS]: The skill fetches code and metadata from well-known services (GitHub and GitLab) based on the specific pull request or repository provided by the user. These network operations are intrinsic to the skill's purpose of analyzing external PRs.
- [PROMPT_INJECTION]: As the skill is designed to process untrusted external data (pull request descriptions, comments, and code diffs), it has an inherent surface for indirect prompt injection. Malicious instructions embedded in a PR could attempt to influence the AI's assessment. However, the instructions do not contain any patterns attempting to bypass platform safety guidelines or override agent behavior.
- [DATA_EXPOSURE]: The skill creates local review artifacts in a
reviews/directory. It does not attempt to access sensitive system files (e.g.,.ssh,.aws) or exfiltrate local data beyond the intended reporting and interaction with the specified Git provider.
Audit Metadata