platform-cli

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates the management of machine learning projects and cloud training runs through the execution of ul and ul cloud CLI commands.
  • [EXTERNAL_DOWNLOADS]: Fetches the official OpenAPI specification from https://platform.ultralytics.com/openapi.json. This download targets a vendor-owned domain to provide the agent with current API request structures.
  • [REMOTE_CODE_EXECUTION]: Scanners detected a command pattern where remote content is piped to python3. Detailed review shows this is a benign operation (curl ... | python3 -c "...") that uses a static, local Python script to parse JSON data from a vendor-owned source rather than executing arbitrary remote code.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user inputs and local files (via @file and @path syntax) which is interpolated into shell commands. This represents a standard capability for a CLI wrapper, though it defines a surface for potential indirect prompt injection.
  • [SAFE]: All identified Python packages (ultralytics, ultralytics-platform) and external domains (ultralytics.com) are official resources belonging to the skill's author.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 03:07 AM
Security Audit — agent-trust-hub — platform-cli