argent-react-native-optimization
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including ESLint JSON reports, profiler analysis outputs, and system logs via
debugger-log-registry. The skill does not define boundary markers or delimiters to isolate this data from instructions. The agent possesses significant capabilities including writing to the filesystem to apply fixes, installing packages vianpm install, and interacting with the device usingargent-device-interact. No sanitization or filtering of these ingestion points is implemented, creating a surface where malicious code or logs could attempt to influence agent operations. - [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the
eslint-plugin-react-perfpackage from the official NPM registry. This is a standard and expected performance utility for the described use case. - [DYNAMIC_EXECUTION]: The workflow incorporates the use of
debugger-evaluateto verify that optimization fixes work as intended. This entails the execution of agent-modified code within the application context.
Audit Metadata