The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the zerion-cli and multiple partner SDKs (e.g., @bankr/cli, @lifi/sdk, @somnia-chain/reactivity) through npm. These installations are appropriate for a crypto wallet management tool and include explicit instructions for supply-chain security, such as requiring a 15-day release-age cooldown for third-party packages to mitigate the risk of compromised fresh releases.
[COMMAND_EXECUTION]: The skill is designed to operate by executing the zerion binary in a shell environment. This is the intended and documented method for the AI agent to interact with the wallet API and the local encrypted vault.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including blockchain RPCs and various DeFi protocol APIs. To mitigate potential injection risks, the instructions emphasize using structured JSON for data exchange and encourage the agent to perform on-chain verification and human-led review before committing to transactions.
[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes platform-specific syntax to dynamically load configuration files (e.g., network-config.json) into the agent's context. This usage is confined to loading documentation and configuration necessary for the skill's operation.
[CREDENTIALS_UNSAFE]: The skill handles sensitive wallet credentials but implements extensive safety measures. These include AES-256-GCM encryption for local keystores, strict file permission requirements (0600) for passphrase files, and a 'Review Threshold' feature that automatically routes transactions above a user-defined USD value to a web-app handoff for manual human signing.

zerion