application-performance-performance-optimization

Pass

Audited by Gen Agent Trust Hub on Jul 7, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill interpolates user-controlled input ($ARGUMENTS) into instructions for various high-capability sub-agents without implementing boundary markers or sanitization. This presents an indirect prompt injection surface.\n
  • Ingestion points: User input from $ARGUMENTS is used across all profiling, optimization, and testing phases (e.g., Steps 1, 4, 7, 10).\n
  • Boundary markers: There are no delimiters or 'ignore-instructions' guards surrounding the interpolated input.\n
  • Capability inventory: Sub-agents (e.g., performance-engineer, database-optimizer, test-automator) are tasked with modifying database configurations, backend code, and CI/CD pipelines (GitHub Actions).\n
  • Sanitization: The workflow does not include steps to validate or sanitize the input before it reaches the sub-agents.\n- [DATA_EXFILTRATION]: The skill involves access to sensitive data types for profiling and bottleneck identification, which is an expected part of its performance optimization functionality.\n
  • Evidence: The instructions include generating heap dumps for memory analysis, tracing I/O operations, and analyzing database slow query logs.\n
  • Context: These actions are scoped to performance profiling and are accompanied by instructions to use established APM tools like DataDog or New Relic.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 7, 2026, 03:02 AM
Security Audit — agent-trust-hub — application-performance-performance-optimization