agency-performance-benchmarker

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest and analyze external performance data, including Real User Monitoring (RUM) metrics and API response payloads. This functionality introduces a surface for indirect prompt injection.
  • Ingestion points: The agent processes external data streams from RUM tools and API endpoints as part of the benchmarking workflow described in SKILL.md.
  • Boundary markers: The instructions lack specific markers or guidance for the agent to differentiate between benchmarking data and potential malicious instructions embedded within that data.
  • Capability inventory: The agent uses results to perform bottleneck analysis and recommend optimizations, which could be influenced by injected instructions.
  • Sanitization: No sanitization or filtering of external performance metrics or API response content is defined in the skill logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 07:56 AM
Security Audit — agent-trust-hub — agency-performance-benchmarker