honey-ccr

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes the eson command-line utility to perform data compression (eson crush) and recovery (eson retrieve).
  • [EXTERNAL_DOWNLOADS]: The skill requires the external tool eson and references concepts from 'headroom's SmartCrusher', which are dependencies not provided within the skill itself.
  • [PROMPT_INJECTION]: The skill introduces an indirect prompt injection surface by automatically modifying tool outputs (PostToolUse hook).
  • Ingestion points: Raw tool output from the Bash environment is processed before reaching the agent context.
  • Boundary markers: Uses a JSON sentinel {"_ccr":"..."} to indicate where data was dropped.
  • Capability inventory: The skill uses eson crush and eson retrieve for data manipulation and file-system access (via the .honey-ccr/ cache).
  • Sanitization: There is no explicit sanitization of the input tool data before it is sampled or cached.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 09:22 PM
Security Audit — agent-trust-hub — honey-ccr