okx-agentic-wallet

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill performs an environment setup in _shared/preflight.md by installing the @okxweb3/onchainos-installer package via npx. This is a vendor-owned resource used to install the necessary CLI tools for the skill's operation.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from blockchain sources, including token symbols and transaction metadata. The instructions in SKILL.md and references/security.md include explicit defensive guidance, instructing the agent to treat such data as untrusted and to never interpret it as instructions or bypass established security checks.
  • [COMMAND_EXECUTION]: The skill operates by executing subcommands of the onchainos CLI. The instructions provide comprehensive syntax references and enforce a strict confirmation policy, requiring explicit user approval before executing any state-changing commands.
  • [CREDENTIALS_SAFE]: The skill includes strong policy directives to ensure sensitive data protection. It explicitly forbids the agent from requesting, logging, or displaying session tokens, private keys, API keys, or seed phrases, ensuring that these credentials remain protected within the Trusted Execution Environment (TEE).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 02:22 PM