The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data through the --metadata, --uri, and --value arguments, which are subsequently stored on-chain. While this represents a potential surface for indirect injection if other agents process this metadata without sanitization, the skill itself performs hex-encoding validation and size limits (512 bytes) on these inputs.
[COMMAND_EXECUTION]: The skill uses the commander library to define its CLI interface and executes blockchain operations through a local service layer. No arbitrary shell command execution or unsafe subprocess spawning was detected.
[CREDENTIALS_UNSAFE]: The skill interacts with the wallet-manager service to sign transactions. It correctly enforces a security model where write operations require an unlocked wallet session, and no sensitive credentials or private keys are hardcoded or exposed.
[SAFE]: The implementation in identity.ts follows secure coding practices, including input normalization for hex strings, error handling for JSON parsing, and clear separation of concerns between the CLI layer and the underlying blockchain service.

identity