data-client-react
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill provides pre-written JavaScript snippets designed for execution in the user's browser via the evaluate_script tool. This includes a debug shim that monkey-patches internal library methods to log action data to a global variable.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a data ingestion surface by allowing the agent to read application state and action logs directly from the browser runtime environment. * Ingestion points: Data is retrieved through evaluate_script calls that return denormalized responses and entity state from the browser context in references/devtools-debugging.md. * Boundary markers: There are no instructions for using delimiters or boundary markers when the agent processes data retrieved from the application store. * Capability inventory: The agent possesses capabilities to execute code in the browser and perform file operations, creating a risk if store data contains malicious instructions. * Sanitization: The skill does not implement or recommend sanitization for the data retrieved from the browser store before it is processed by the agent.
Audit Metadata