new-session
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell scripts and Git commands to manage local and remote repository state, including stashing changes, fetching updates, and managing branches.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from GitHub APIs (PR branch names) and Git metadata. While this represents a potential injection surface, the data is used for state verification and naming availability checks rather than being directly interpreted as agent instructions.
- [DYNAMIC_EXECUTION]: The handoff verification script executes an inline Python snippet to perform URL encoding of branch names. This is a standard utility function and does not process untrusted code.
Audit Metadata