cx-cases
Pass
Audited by Gen Agent Trust Hub on Aug 13, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted external data from case event timelines and human-generated comments, creating an attack surface for indirect prompt injection.
- Ingestion points: Case comments and alert metadata retrieved via
cx cases events listandget_alerts_object. - Boundary markers: Absent. The instructions lack specific delimiters or directions for the agent to ignore instructions embedded within case telemetry or comments.
- Capability inventory: The skill utilizes the
cxCLI tool for case lifecycle actions such as assignment, acknowledgment, and resolution. - Sanitization: Absent. No filtering, validation, or escaping of ingested external content is specified.
- [COMMAND_EXECUTION]: Operations are performed using the vendor-provided
cxCLI tool for case management, which is consistent with the skill's primary purpose.
Audit Metadata