skills/loonghao/vx/vx-best-practices/Gen Agent Trust Hub

vx-best-practices

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface
  • Ingestion points: The skill describes workflows where agents read project configuration files (vx.toml), project source code via search tools (rg), and external metadata via CLI tools (gh, git).
  • Boundary markers: No explicit delimiters or instructions are provided to the agent to disregard instructions embedded within these external data sources.
  • Capability inventory: The skill assumes the ability to execute shell commands via the vx utility, which wraps package managers and build tools (npm, cargo, go).
  • Sanitization: No specific sanitization or validation of content from external sources is documented.
  • [COMMAND_EXECUTION]: Documents the use of Starlark-based 'providers' (provider.star) to define tool installation and execution logic. This involves the execution of scripts to manage tool environments.
  • [EXTERNAL_DOWNLOADS]: References the use of the loonghao/vx@main GitHub Action for integrating tool management into automated CI/CD workflows.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 04:14 AM
Security Audit — agent-trust-hub — vx-best-practices