wuji-cli-user

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill interacts with the local environment exclusively through the wuji CLI binary to manage user profiles and calibration data. These commands are local and do not initiate network or hardware connections.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by allowing the import of external data bundles. 1. Ingestion points: Data enters the system via the wuji user import <path> command specified in SKILL.md. 2. Boundary markers: The instructions mandate the use of a --preview flag and require the agent to manually relay the bundle's source, version, and content details to the user for verification before execution. 3. Capability inventory: The skill can create, modify, and delete profile data and directories within the ~/.wuji/sdk directory using the provided CLI tool. 4. Sanitization: The skill relies on the CLI's internal bundle version validation and an explicit requirement for human-in-the-loop confirmation before performing destructive imports.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 07:55 PM
Security Audit — agent-trust-hub — wuji-cli-user