The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interacts with the onchainos CLI tool using execFileSync to fetch market data and prepare transactions. This is the core functionality of the skill. The instructions explicitly warn the agent to treat all CLI output as untrusted external content and require the user to manually install the CLI rather than attempting automated installation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from the OKX CLI and Helius APIs. It follows security best practices by implementing a 'trust but verify' model, requiring explicit human-in-the-loop confirmation before any transaction is signed or submitted, and performing automated safety checks on token data (e.g., honeypot detection).
[DATA_EXFILTRATION]: The skill manages sensitive information including Solana private keys and API credentials for Helius and OKX. It correctly follows secret management best practices by using environment variables and ensuring that transaction signing occurs locally before submission to the Helius Sender endpoint. No suspicious network exfiltration patterns were identified.
[EXTERNAL_DOWNLOADS]: The skill references several external resources such as the Helius MCP server and the OKX skill library. It provides safe guidance by directing the user to official GitHub repositories and NPM packages from well-known services, and it explicitly instructs the AI not to execute remote installation scripts on behalf of the user.

okx