The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill configuration instructs the agent to download and run the @bnb-chain/mcp package from the official npm registry at runtime using npx.
[COMMAND_EXECUTION]: The connection instructions utilize npx to execute the MCP server, which involves running shell commands on the host machine to start the service.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes untrusted external data from the blockchain and storage networks.
Ingestion points: External data enters the context through tools such as read_contract, get_erc20_token_info, get_nft_info, and gnfd_get_object_info (found in references/evm-tools-reference.md and references/greenfield-tools-reference.md).
Boundary markers: The instructions in SKILL.md include explicit safety requirements for the agent to confirm the recipient, amount, and network with the user before executing any write operations.
Capability inventory: The skill has access to high-privilege capabilities including native/ERC20/NFT transfers, smart contract writes, and Greenfield storage management (file uploads/downloads).
Sanitization: There are no explicit instructions for the sanitization of incoming data from blockchain calls before it is presented to the user or used in subsequent logic.

bnbchain-mcp