sota-performance
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, including source code, database execution plans, and performance telemetry, which creates a potential surface for indirect prompt injection. Ingestion points: The agent reads source code, shell output from profiling tools, and database diagnostic data during the AUDIT phase (e.g., in rules/02-algorithms-data-structures.md and rules/05-caching.md). Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded content within the files being audited. Capability inventory: The skill utilizes file system search tools (grep, find), process monitoring tools (ps, perf, strace), and database query analysis (EXPLAIN ANALYZE). Sanitization: No explicit sanitization or filtering of the processed content is specified before analysis.
- [SAFE]: The skill incorporates strong security-conscious performance advice, such as warnings against timing attacks, unauthorized data access through shared caches, and accidental exposure of debug endpoints (rules/01-methodology.md, rules/05-caching.md).
Audit Metadata