The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions are purely technical and educational. There are no attempts to override agent constraints, bypass safety filters, or use adversarial patterns like DAN or instruction disregard.
[DATA_EXFILTRATION]: No sensitive file access or exfiltration patterns were detected. The contract examples use placeholder IPFS hashes for metadata and standard internal storage variables.
[COMMAND_EXECUTION]: The skill contains smart contract code and markdown but no shell commands, subprocess calls, or privilege escalation attempts. Access control is managed through the standard OpenZeppelin 'Ownable' pattern.
[EXTERNAL_DOWNLOADS]: Smart contract dependencies are limited to well-known, trusted libraries (OpenZeppelin and ERC721A). These are standard in the Ethereum ecosystem and do not represent a security risk.
[SAFE]: The use of Base64 encoding in the 'OnChainNFT' example is a standard technique for generating on-chain SVG and JSON metadata for NFTs. It is not used to obfuscate malicious payloads.

nft-standards