create-github-issue
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
ghandgitcommand-line tools to interact with the repository and GitHub services. This includes reading status, logs, and creating issues, which are standard operations for its stated purpose. - [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting and strictly following the structure of YAML files found in
.github/ISSUE_TEMPLATE/and.github/labels.json. A malicious repository could theoretically include instructions within these files to manipulate agent behavior. - Ingestion points:
.github/ISSUE_TEMPLATE/*.ymland.github/labels.json(SKILL.md). - Boundary markers: None. The skill instructs the agent to follow the template structure "exactly" and "without exception."
- Capability inventory: Shell command execution (
gh,git) and file system writes (writetool). - Sanitization: The skill does not describe any validation or sanitization steps for the data retrieved from repository files.
- [COMMAND_EXECUTION]: The skill includes an instruction to avoid asking for user confirmation before creating issues when intent is clear. While intended for efficiency, this increases the potential impact of indirect prompt injection if the agent is manipulated by malicious repository data.
Audit Metadata