profiling-performance
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions to ingest and analyze performance profile data generated from external or local web applications. This creates a surface where a malicious web page could theoretically attempt to influence the agent's analysis by including deceptive function names or metadata within the profiling logs. However, this risk is inherent to the task of performance profiling and is considered part of the skill's intended operational context.
- Ingestion points: CPU profile data in JSON and Markdown formats are read from the
~/.cursor/browser-logs/directory. - Boundary markers: The instructions do not define specific delimiters or instructions to the agent to treat the profile data as untrusted content.
- Capability inventory: The skill utilizes browser automation tools (
browser_navigate,browser_click,browser_fill) and standard file system read capabilities. - Sanitization: There is no explicit sanitization or validation performed on the profile data before it is processed by the agent.
Audit Metadata