token-optimization

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is purely instructional and provides strategies for context management, such as prompt caching and conversation history compression, without introducing malicious code or unsafe tools.
  • [PROMPT_INJECTION]: The instructions explicitly direct the agent to 'never prune system instructions, safety headers, or compliance text' during optimization tasks, which actively prevents accidental context-loss that could lead to safety bypasses.
  • [DATA_EXFILTRATION]: No sensitive data access or network-based exfiltration patterns were found. Mentions of local files like ~/.claude/settings.json are used only to describe standard configuration patterns for users.
  • [REMOTE_CODE_EXECUTION]: There is no evidence of remote script fetching or execution. References to common CLI tools (e.g., git, gh, aws) are used within technical documentation as recommended workflows for developers.
  • [COMMAND_EXECUTION]: While the skill discusses CLI usage and tool-loop patterns, it does not provide any scripts or automated commands that perform these actions on the host system.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 02:32 PM
Security Audit — agent-trust-hub — token-optimization