make-a365-agent

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Remote Script Execution: The skill provides instructions to download and execute the Microsoft Dev Tunnel CLI installation script (https://aka.ms/DevTunnelCliInstall) using curl | bash on Linux systems. This is a common pattern for CLI tool distribution from established vendors, though it bypasses local package manager verification.
  • Command Argument Interpolation: The skill ingests user-defined inputs, such as agent names and project directories, which are then interpolated into shell commands (e.g., cd "<project_dir>", a365 setup all --agent-name <agent_name>).
  • Ingestion points: Inputs for agent_name, project_dir, and messagingEndpoint are collected from the user or local config files.
  • Boundary markers: The instructions include specific naming rules (e.g., letters, numbers, and hyphens only) to guide the agent, but do not employ explicit boundary delimiters for the processed values.
  • Capability inventory: The skill utilizes the Bash tool to execute various configuration and setup commands.
  • Sanitization: The skill instructs the agent to enforce character restrictions on the agent name and uses double-quoting for directory paths, which serves as a basic mitigation against command injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 09:55 PM