okx-agentic-wallet

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill includes instructions for downloading an installation script (install.sh) and checksums to ensure the onchainos CLI tool is present. This is a standard operational step for a CLI-integrated wallet skill from a trusted vendor.
  • [REMOTE_CODE_EXECUTION]: The skill describes an automated installation and update mechanism via shell scripts when the required CLI tool is missing or out of date. This behavior is consistent with the intended functionality of the OKX Agentic Wallet integration and targets the vendor's own binary.
  • [COMMAND_EXECUTION]: The skill's primary functionality is achieved by executing various shell commands via the onchainos CLI to manage wallets, perform swaps, and interact with smart contracts.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes untrusted data from the blockchain. 1. Ingestion points: Token names, symbols, and on-chain metadata retrieved via the CLI. 2. Boundary markers: The instructions mandate treating all CLI output as untrusted external content. 3. Capability inventory: Shell command execution for wallet transfers and contract interactions. 4. Sanitization: The skill contains explicit rules for the agent to refuse requests to extract credentials or bypass checks and to treat data as factual only, avoiding the interpretation of data as instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 04:06 PM