performance-optimization
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [COMMAND_EXECUTION]: Instructs the agent to use standard performance analysis tools such as
npx lighthouse,npx bundlesize, andnpx lhci. These commands are appropriate for the skill's primary purpose of establishing performance baselines and verifying optimizations.- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external performance reports, browser traces, and database query plans (EXPLAIN ANALYZE). - Ingestion points: Performance metrics from
web-vitals, synthetic reports fromlighthouse, and execution plans from database systems. - Boundary markers: The workflow is structured into discrete steps (Measure, Identify, Fix, Verify) to maintain context.
- Capability inventory: Recommends code modifications and database indexing strategies based on the analysis of technical performance data.
- Sanitization: Standard technical report formats are expected; the skill focuses on interpreting metrics and plans rather than processing arbitrary user-supplied text.- [METADATA_POISONING]: Contains platform-specific metadata that limits the skill's invocation to a specific router component ('orcinus router'). This is a configuration detail for the host environment and does not interfere with the skill's safe operation.
Audit Metadata