add-dial-tool

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes several local commands on the host to manage the installation, including ncl for agent group management and onecli for secret vaulting. All commands are purpose-built for the skill's stated function of integrating the Dial service.
  • [EXTERNAL_DOWNLOADS]: The skill downloads and installs the official @getdial/cli package from the NPM registry. These downloads originate from the vendor's official package repository and are pinned to specific versions, which is a standard and safe practice for tool integration.
  • [CREDENTIALS_UNSAFE]: The skill handles an API key from a local file (dial/auth.v1.json) to move it into a secure vault. It uses a temporary file with restricted permissions (0600) and ensures the secret is never passed via command-line arguments (argv), preventing exposure in process lists or logs.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes agent IDs and emails provided by the operator. It implements strict validation patterns (regex) for these inputs and uses boundary markers in its shell scripts to prevent command injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 06:59 PM