ccbox-insights
Pass
Audited by Gen Agent Trust Hub on Feb 22, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [Data Exposure] (LOW): The skill accesses session logs via the
ccboxCLI. These logs are a source of sensitive information as they may contain environment variables, API call details, and proprietary code from previous agent interactions. Accessing this data is inherent to the skill's primary purpose. - [Indirect Prompt Injection] (LOW): The skill processes untrusted historical log data which could contain malicious instructions embedded by an attacker in a previous session. * Ingestion points:
ccbox historyandccbox sessions(SKILL.md). * Boundary markers: None identified in the workflow to separate log data from agent instructions. * Capability inventory: Shell execution (ccbox) and file system access. * Sanitization: No sanitization of log content is performed before processing. - [Command Execution] (SAFE): The skill requires and uses the
ccboxCLI tool. The commands used (projects,sessions,history) are consistent with the skill's stated goal of log analysis and do not exhibit malicious behavior.
Audit Metadata