woostack-change
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied goals to implement code changes, creating a surface where malicious input could influence the agent's implementation or documentation.
- Ingestion points: The
<goal>parameter in the/woostack-change <goal>command. - Boundary markers: The instructions mandate an admission phase to clarify the scope and outcome before any mutation occurs, explicitly listing non-goals and rerouting criteria.
- Capability inventory: The skill uses
gitandGraphitefor repository management, including worktree creation and PR submission, and allows for file system modifications to implement changes. - Sanitization: The skill explicitly instructs the agent to validate and narrow untrusted inputs when touching inter-application boundaries.
- [COMMAND_EXECUTION]: The skill utilizes local shell commands for version control and workflow management.
- Evidence: Extensive use of
git(e.g.,git worktree add,git diff) andGraphite(e.g.,gt track --parent,gt submit) commands to manage isolated workspaces and deliver pull requests.
Audit Metadata