ln-812-optimization-researcher

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions specify a fallback to fetch shared documentation and reference files from the author's GitHub repository (github.com/levnikolaevich/claude-code-skills) if local directories are missing. This ensures the availability of required research templates and tool priority chains.\n- [COMMAND_EXECUTION]: The skill utilizes grep to perform discovery within the local codebase, searching for performance-related infrastructure such as caching, connection pooling, and batch processing methods to assess optimization feasibility.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests data from external web searches and local file content to generate optimization reports.\n
  • Ingestion points: WebSearch results and local source code analyzed via grep.\n
  • Boundary markers: None specified for external search data.\n
  • Capability inventory: Shell command execution (grep) and writing performance artifacts to the file system.\n
  • Sanitization: No explicit sanitization of search results is defined.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 06:29 AM