skills/okx/onchainos-skills/okx-guide/Gen Agent Trust Hub

okx-guide

Fail

Audited by Gen Agent Trust Hub on Sep 24, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill utilizes npx to download and execute the @okxweb3/onchainos-installer package in references/install-update.md and _shared/preflight.md. This package is a vendor-managed utility for environment setup and version updates.\n- [COMMAND_EXECUTION]: Several files in the skill invoke the onchainos CLI to perform operations such as checking wallet status, retrieving balances, verifying geographical restrictions, and managing agent-related metadata via subcommands like agent get-my-agents, agent search, and agent task-in-progress.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from external marketplace and task systems that may be influenced by third-party inputs.\n
  • Ingestion points: External data is ingested through the outputs of agent search (marketplace provider data) and agent task-in-progress (task titles and descriptions) in references/registered-home.md.\n
  • Boundary markers: The skill lacks explicit delimiters or boundary markers to separate untrusted external content from system instructions when presented to the agent context.\n
  • Capability inventory: The agent has access to the onchainos CLI for wallet and task management, as well as the ability to route to other ecosystem skills like okx-defi and okx-dex-market.\n
  • Sanitization: The instructions in references/registered-home.md explicitly mandate treating all external fields as untrusted and specifically prohibit the exposure of signing addresses to the user, providing a layer of data exposure protection.
Recommendations
  • CRITICAL: 2 file(s) identified as malware by FileRep - DO NOT USE
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Sep 24, 2026, 01:19 PM
Security Audit — agent-trust-hub — okx-guide