The Agent Skills Directory

Command Execution Capability: The skill provides documentation and code examples for the InvokeAgentRuntimeCommand API, which allows for shell command execution within an agent's session environment. While this is a powerful capability, the documentation includes prominent security warnings regarding command injection and provides advice on validation and IAM scoping.
Handling of Sensitive Credentials: The skill facilitates the setup of payment integrations which require provider-specific secrets (e.g., Coinbase or Stripe API keys). It provides instructions for manual credential entry using the AgentCore CLI and notes that these secrets are stored in a local .env.local file. It also recommends the use of "QuickCreate" for Coinbase to minimize direct handling of secrets.
Network Operations and Data Ingestion: The inclusion of the x402_fetch tool allows agents to make network requests to external URLs and process their responses, specifically for handling micropayments. The tool implements basic security checks, such as enforcing HTTPS and blocking access to private or internal network addresses (SSRF protection).
Integration with Third-Party Services: The skill provides guidance on integrating with services like Privy for payment delegation, which involves downloading and running a reference frontend from a public GitHub repository. This is presented as a developer-led configuration step.

agents-build