auditing-performance
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute bundle analysis tools using
npx(e.g.,@next/bundle-analyzerandvite-bundle-visualizer). These are standard diagnostic tools for web development. - [INDIRECT_PROMPT_INJECTION]: The skill processes application source code and database schemas to identify performance bottlenecks. While this involves ingesting untrusted external data, the risk is inherent to the auditing task and the skill does not grant excessive privileges.
- Ingestion points: Project source code,
package.json, and database schema files (SKILL.md). - Boundary markers: None explicitly defined in instructions.
- Capability inventory: Limited to shell execution of bundle analyzers and file system reads.
- Sanitization: None specified for the analysis process.
Audit Metadata