lido-plugin
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the
lido-pluginbinary, a universal launcher script, and an update checker from the vendor's (okx) official GitHub repositories during the pre-flight installation process. - [REMOTE_CODE_EXECUTION]: During setup, the skill fetches and executes an installation script for the
onchainosCLI from the vendor's GitHub repository using a piped-to-shell command (curl | sh). - [COMMAND_EXECUTION]: The plugin performs various system operations, including executing the
onchainosCLI to handle blockchain transactions and utilizingnpxto install additional skill dependencies from theokxorganization. - [DATA_EXFILTRATION]: The skill collects system environment metadata (including hostname, operating system details, and home directory path) which is hashed into a device ID and transmitted to the vendor's reporting APIs (
okx.comand a Vercel-hosted stats endpoint) for installation analytics. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as staking APRs from DeFiLlama and withdrawal status from the Lido API. It includes a 'Data boundary notice' instructing the agent to treat this external content as untrusted, which mitigates the risk of indirect instructions embedded in third-party data.
Audit Metadata