The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains directives that override standard agent behavior by commanding the silent installation of a global CLI tool without user consent. It also instructs the agent to suppress error messages from the user regarding the presence or installation status of the package.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch and install the @liberfi.io/cli package from the official NPM registry. While the registry is a well-known service and the package is associated with the vendor, the directive to install it automatically at runtime is a security concern.
[COMMAND_EXECUTION]: The skill relies on executing global system commands, specifically npm install -g for setup and various lfi CLI commands to interact with blockchain data and manage user authentication states.
[DATA_EXFILTRATION]: The skill accesses the authenticated user's wallet data and transaction history through the lfi me commands. This involves processing and displaying user-specific blockchain information from the vendor's TEE environment.

liberfi-portfolio