context-optimization

Pass

Audited by Gen Agent Trust Hub on Apr 25, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes local CLI tools hermes and rtk to perform analytics and session management.
  • [INDIRECT_PROMPT_INJECTION]: The skill involves reading output from session and token analytics tools to update rules in CLAUDE.md. This represents an indirect prompt injection surface where data from tool outputs influences the agent's future instructions. Ingestion points: Command output from hermes insights and rtk gain. Boundary markers: Absent. Capability inventory: Shell execution and file system modification (patching CLAUDE.md). Sanitization: None identified.
  • [PERSISTENCE_MECHANISMS]: The skill includes instructions to configure a cron job (hermes sessions prune) for automated session maintenance. This persistence is legitimate as it aligns with the skill's primary purpose of session maintenance and context optimization.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 25, 2026, 12:08 PM
Security Audit — agent-trust-hub — context-optimization