skills/bofai/skills/agent-wallet/Gen Agent Trust Hub

agent-wallet

Pass

Audited by Gen Agent Trust Hub on Apr 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the @bankofai/agent-wallet package from the official npm registry. This package is provided by the skill's author, BofAI, to enable its core functionality.
  • [COMMAND_EXECUTION]: The skill relies on the agent-wallet CLI for all blockchain operations. It enforces a safety protocol requiring the agent to run --help for every command to verify parameters before execution.
  • [CREDENTIALS_UNSAFE]: The instructions direct the agent to ask the user for a master password in plain text and pass it as a command-line argument. This allows the password to appear in the clear within the chat conversation history and the agent's internal logs.
  • [COMMAND_EXECUTION]: The skill implements a 'least privilege' design for the agent by identifying specific high-risk commands—such as removing wallets or changing passwords—as 'agent-restricted'. The agent is instructed to provide these commands to the user to run in their own terminal rather than executing them directly.
  • [PROMPT_INJECTION]: The skill creates an ingestion surface for indirect prompt injection by signing arbitrary JSON transaction payloads and message strings provided by the user. The skill mitigates this risk by limiting its scope to signing operations and providing clear warnings about key safety.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 22, 2026, 10:23 AM
Security Audit — agent-trust-hub — agent-wallet