submit-feedback
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the GitHub CLI (
gh) to interact with repositories, specifically for listing, viewing, creating, and editing issues and pull requests. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external, potentially attacker-controlled sources and possesses capabilities that could be influenced by malicious content within that data.
- Ingestion points: The skill retrieves external content using
gh issue list,gh pr list, andgh issue viewin steps 4 and 6. - Boundary markers: There are no specific instructions or delimiters used to isolate the external content or instruct the agent to ignore instructions embedded within the issues or PRs it reads.
- Capability inventory: The skill uses the
ghtool for network-based repository management (creating/editing/commenting) and performs file system operations such as creating temporary files withmktempand writing to them via shell redirection. - Sanitization: While the skill correctly uses heredocs (
<<'EOF') and the--body-fileflag to prevent shell-level command injection when publishing, it does not provide mechanisms to sanitize the content for potential natural language instructions that could affect agent behavior.
Audit Metadata