triage-issue
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- Command Construction Consideration: The skill uses user-provided arguments ($ARGUMENTS) directly within shell command templates (e.g., gh issue view). While this enables the core functionality of triaging specific issues, it creates a surface where input containing shell metacharacters could potentially influence the execution environment.
- Indirect Prompt Injection Surface: The agent is instructed to read and process the body and comments of GitHub issues, which are external and untrusted data sources. Because the skill then performs write actions—such as closing issues or updating labels—based on its analysis of this content, there is a possibility that instructions embedded within an issue could influence the agent's behavior.
- Ingestion points: GitHub issue content (body and comments) retrieved via the 'gh issue view' command in SKILL.md.
- Boundary markers: None present; the instructions do not explicitly tell the agent to ignore or isolate instructions found within the issue text.
- Capability inventory: The skill utilizes 'gh issue close' and GitHub API POST/DELETE methods to modify repository labels.
- Sanitization: There are no defined sanitization or validation steps for the retrieved issue content before processing.
- External Resource Retrieval: The instructions direct the agent to fetch URLs mentioned in issue comments and bodies to verify their status. This involves interacting with external sites determined by the content of the issue, which can include non-vendor domains and external references.
Audit Metadata