The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill decrypts and parses data from external entity payloads, creating an untrusted ingestion surface. * Ingestion points: Data is processed in the readEncryptedNote function in references/sdk-payload.md. * Boundary markers: No programmatic delimiters are used; the skill relies on instructional warnings to the agent. * Capability inventory: Access to blockchain writes (createEntity) and cryptographic primitives. * Sanitization: The parseNote function enforces a strict JSON schema and a 10,000-character length limit on the decrypted field.
[EXTERNAL_DOWNLOADS]: The skill requires external packages and references documentation. * Installs arkiv-encryption@0.1.0 (vendor resource) and dependencies @arkiv-network/sdk and viem. * References documentation at github.com/SantiagoDevRel/arkiv-encrypted-entities.

arkiv-encryption