clix-user-management

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Prompt Injection (LOW): The skill demonstrates a surface for indirect prompt injection as it processes user-provided configuration files to define execution plans.
  • Ingestion points: The skill reads .clix/user-plan.json or user-plan.json via the validate-user-plan.sh script.
  • Boundary markers: Absent; the agent is instructed to use the plan directly for implementation workflow.
  • Capability inventory: The skill executes local shell scripts and Python/Node code to validate schemas and manage user property logic.
  • Sanitization: The validation script (validate-user-plan.sh) enforces a strict JSON schema, primitive type checks, and regex validation for property keys to prevent arbitrary instruction injection through the schema keys.
  • Dynamic Execution (SAFE): The skill uses a Python heredoc within a bash script for validation. Since the Python code is static and only parameters (file paths) are dynamic, this does not constitute unsafe dynamic code generation from untrusted sources.
  • Data Exposure (SAFE): No patterns of hardcoded credentials, sensitive file access (e.g., SSH keys), or network exfiltration were identified.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:31 PM