github-delivery
Pass
Audited by Gen Agent Trust Hub on Nov 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to manage GitHub repository states using Git commands and GitHub APIs. This includes branch lifecycle management, pull request synchronization, and the reconciliation of repository branch protection rulesets. These operations are functionally scoped to the skill's purpose and include pre-flight verification of commit hashes.
- [EXTERNAL_DOWNLOADS]: The skill incorporates a verification step that fetches publication artifacts from external registries and providers (e.g., package registries, cloud provider APIs) to confirm the success and identity of a release. These downloads target well-known technology services and are used exclusively for integrity verification.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from GitHub Issues and Pull Requests, creating an indirect prompt injection surface. 1. Ingestion points: Data is pulled from issue/PR titles and bodies (SKILL.md). 2. Boundary markers: No explicit delimiters are specified in the instructions. 3. Capability inventory: The skill can perform branch/issue/PR mutations and merge code. 4. Sanitization: No specific input sanitization is mentioned. However, the risk is effectively mitigated by the skill's architectural requirement for explicit human authorization for all 'Release-class' merges to the main branch and its reliance on immutable commit SHAs for state validation.
Audit Metadata