The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides logic for handling external blockchain data and user-supplied naming strings, which are potential vectors for indirect injection. It mitigates these risks by mandating strict validation procedures.
Ingestion points: Untrusted data enters the agent's context through ENS names provided in search inputs, address books, and reverse records retrieved from blockchain resolvers (documented in SKILL.md).
Boundary markers: The instructions specify the use of normalization at all input boundaries and require forward-verification for reverse resolution to ensure data integrity.
Capability inventory: The skill utilizes standard blockchain client libraries (viem, wagmi, ethers); it does not perform arbitrary subprocess calls, file writes, or unsafe network operations.
Sanitization: Explicitly instructs developers to use the @adraffy/ens-normalize library to sanitize and validate names before processing them.
[OBFUSCATION]: The text includes an example of a homoglyph string ('аpple.eth') within an educational context. It serves to illustrate the importance of normalization in preventing phishing and is not used as a functional link or part of an attack.
[SAFE]: The skill incorporates security best practices, including using official documentation (docs.ens.domains) as the source of truth, avoiding hardcoded contract addresses, and preferring fresh RPC data over potentially stale caches for sensitive transactions.

ens