app-store-review
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and evaluate untrusted user code and configuration files, which presents a surface for indirect prompt injection. The instructions include a workflow where the agent may generate shell commands for the user to execute.
- Ingestion points: The agent reads project source code (Swift, Objective-C, TypeScript) and sensitive configuration files such as
Info.plistandapp.jsonin the user's local environment. - Capability inventory: The instructions guide the agent to suggest using the
ghcommand-line tool (gh issue create) to submit feedback to the maintainer's repository. - Boundary markers: The skill body does not specify explicit boundary markers or delimiters for the user-supplied code it processes.
- Sanitization: The instructions provide robust guidelines for data minimization, explicitly directing the agent to exclude user code, proprietary details, and credentials from any drafted feedback unless the user manually intervenes to include them.
Audit Metadata