ccbox-insights

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The required workflow uses local ccbox CLI commands to read local session logs (ccbox history, ccbox sessions), which record previous interactions including user messages and tool outputs. Since these session logs can contain arbitrary text/prompts entered by users during past sessions, reading them introduces indirect prompt injection risk from an untrusted source.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skills installation command references an external GitHub repository (diskd-ai/ccbox) which constitutes an unverifiable operational dependency during setup, classified under high tier.