The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes external data from the Circle Gateway API and user-supplied parameters.
Ingestion points: Attestations from the Circle Gateway API (e.g., in 'references/evm-to-evm.md') and user-supplied transaction parameters like recipients and amounts.
Boundary markers: The implementation lacks explicit boundary markers or 'ignore' instructions when interpolating this external data into the agent's context.
Capability inventory: The skill is capable of constructing, signing, and submitting financial transactions across multiple blockchain networks.
Sanitization: While standard blockchain formatting utilities are used, the skill does not implement high-level content validation for attestations received from external APIs.
[EXTERNAL_DOWNLOADS]: The skill depends on several third-party libraries for blockchain connectivity and vendor-specific services.
Node.js dependencies: Packages such as 'viem', 'wagmi', and '@solana/web3.js' are required, alongside the official vendor SDK '@circle-fin/developer-controlled-wallets'.
[DATA_EXFILTRATION]: The skill performs network requests to communicate with the Circle Gateway API and various public blockchain RPC nodes.
External endpoints: It references public RPC services, including the Hyperliquid EVM testnet RPC ('https://rpc.hyperliquid-testnet.xyz/evm'), which was flagged by automated scans but is a standard endpoint for that network.
Credential handling: The skill utilizes environment variables for sensitive credentials (e.g., 'CIRCLE_API_KEY') and includes explicit rules to prevent hardcoding or logging these secrets.

use-gateway