token-optimization
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of documentation and reference implementations for optimizing AI agent costs. No malicious patterns, unauthorized data access, or persistence mechanisms were found.
- [INDIRECT_PROMPT_INJECTION]: The skill defines logic for processing external prompts through a
PromptCompressorandComplexityClassifier. While these components handle untrusted input, their functionality is restricted to string manipulation (whitespace removal, instruction truncation) and routing, with no execution of the processed content. - [EXTERNAL_DOWNLOADS]: Includes links to the vendor's GitHub repository and production platform for context. These are static documentation links and do not perform any automated downloads or script executions at runtime.
Audit Metadata