triage
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access were detected in the skill files.
- [PROMPT_INJECTION]: The instructions do not contain attempts to override safety protocols or bypass agent constraints. The workflow for processing external issue data includes human oversight, requiring maintainer confirmation before transitions or actions, which mitigates indirect prompt injection risks.
- [DATA_EXFILTRATION]: The skill reads issue metadata and local markdown files in the
.out-of-scope/directory. This data access is limited to the project's own repository and is used for its intended purpose of triage context. - [REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of remote scripts, runtime code generation, or unsafe deserialization.
Audit Metadata