performance

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to audit and optimize external web pages via URLs. This processing of untrusted external content represents a surface for indirect prompt injection if the target pages contain malicious instructions.
  • Ingestion points: The agent accesses external content through tools like performance_start_trace and lighthouse_audit as described in SKILL.md and references/MEASUREMENT.md.
  • Boundary markers: The instructions do not specify the use of delimiters or specific commands to ignore potential instructions embedded within the analyzed page data.
  • Capability inventory: The skill possesses the capability to perform detailed performance analysis and suggests modifications to the site's source code based on these findings.
  • Sanitization: While the references/RUM.md file advises against collecting personal data (PII) in monitoring payloads, there is no explicit sanitization step for the data ingested during the auditing process.
  • [EXTERNAL_DOWNLOADS]: The skill references several external resources and tools for performance measurement.
  • The references/MEASUREMENT.md file suggests the use of the Lighthouse CLI for local performance audits.
  • The references/RUM.md file provides implementation examples using the web-vitals JavaScript library.
  • All referenced tools and libraries are official packages maintained by the Google Chrome team and are standard in the web development industry.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 12:40 AM
Security Audit — agent-trust-hub — performance