perf-code-paths
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFENO_CODE
Full Analysis
- Prompt Injection (SAFE): The instructions focus on performance analysis and do not contain attempts to override agent behavior or bypass safety filters.
- Data Exposure & Exfiltration (SAFE): No hardcoded credentials or network exfiltration patterns were detected.
- Unverifiable Dependencies & Remote Code (SAFE): No external packages or remote scripts are referenced or executed.
- Indirect Prompt Injection (SAFE): While the skill instructs the agent to analyze local source code (an ingestion point), it does not include high-risk capabilities like file writing or network requests. 1. Ingestion points: Local repository files accessed via grep or repo-map. 2. Boundary markers: Absent. 3. Capability inventory: Read-only file analysis. 4. Sanitization: Absent.
Audit Metadata