react-native
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to perform automated discovery of installed skills by running 'ls' on hidden platform directories (e.g., ~/.claude/skills/) to facilitate integrated tool recommendations.
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it is designed to ingest and analyze untrusted project data (source code and configuration files) while maintaining access to shell tools (npm, npx, eas) without explicit boundary markers or input sanitization.
- Ingestion points: Local filesystem access to project source code, package manifests, and Expo configuration files (app.json/app.config.ts).
- Boundary markers: Absent; the skill instructions do not specify delimiters to isolate external project content from the instruction stream.
- Capability inventory: Execution of shell commands via npm, npx, and eas CLI tools for project initialization, dependency management, and cloud builds.
- Sanitization: None; the skill does not provide logic to validate or filter the content of analyzed project files.
Audit Metadata