okx-buildx-hackathon-agent-track

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill facilitates environment setup by fetching and executing installation scripts (install.sh) from the vendor's official GitHub repositories (okx/onchainos-skills and okx/plugin-store). These operations are part of the intended project configuration.
  • [EXTERNAL_DOWNLOADS]: Automated scripts download reference documentation, configuration manifests, and skill dependencies from trusted domains including web3.okx.com, docs.uniswap.org, and moltbook.com.
  • [COMMAND_EXECUTION]: The skill employs shell commands for local project management, including an automated version-checking script that ensures the agent is using the latest version of the hackathon instructions.
  • [PROMPT_INJECTION]: As the skill interacts with a public feed on Moltbook, it acknowledges the risk of indirect prompt injection from third-party content. It mitigates this by instructing the agent to treat all external submission data as untrusted and to prioritize its internal scoring guidelines over embedded instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 12:20 AM