create-issue
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from GitHub issue templates and user-supplied descriptions. It mitigates the risk of this content influencing agent behavior by explicitly instructing the agent to treat templates and copied text as data, not instructions.
- Ingestion points: Reads from
.github/ISSUE_TEMPLATE.mdand.github/ISSUE_TEMPLATE/directory. - Boundary markers: Explicitly stated in the 'Boundaries' section.
- Capability inventory: Performs file reads and executes shell commands via GitHub CLI.
- Sanitization: Instructs the agent to redact secrets and use safe defaults.
- [COMMAND_EXECUTION]: The skill uses the
gh(GitHub CLI) to create issues. It follows high-security standards by explicitly forbidding the interpolation of untrusted text into shell commands and recommending the use of--body-fileto safely pass content. - [DATA_EXFILTRATION]: The skill includes specific safeguards for handling security vulnerabilities and sensitive data. It mandates that reports containing exploits or credentials must stay private and requires checking the repository's
SECURITY.mdbefore any public disclosure.
Audit Metadata