git-hosting
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes official GitHub (
gh) and GitLab (glab) command-line interfaces for repository management, which is appropriate for its intended functionality. - [INDIRECT_PROMPT_INJECTION]: The skill mitigates risks associated with processing external repository data by providing clear security boundaries:
- Ingestion points: External data is ingested from issue/PR titles, bodies, and comments via API responses.
- Boundary markers: It recommends using body files (
--body-file) and structured arguments to isolate multiline user content. - Capability inventory: Actions include repository mutations, setting changes, and branch protection updates via authenticated CLI/API tools.
- Sanitization: It warns against using shell interpolation for serialization and advises following native request schemas.
- [PRIVILEGE_ESCALATION]: The documentation covers high-privilege operations such as modifying branch protection and repository settings, providing guidelines to respect existing administrative constraints and verify the state after each mutation.
- [DATA_EXFILTRATION]: Includes proactive instructions to prevent the leakage of sensitive information, such as avoiding the inclusion of tokens in command arguments or printed output, and following rate-limiting headers for safe transport.
Audit Metadata