The Agent Skills Directory

[SAFE]: The skill facilitates legitimate interaction with the GL-36 protocol on the Solana blockchain. It utilizes well-known services (Solana RPC) and vendor-specific libraries (@gl-36/sdk) that align with the provided author context. All external references and dependencies are appropriate for the skill's stated purpose of managing NFT collections.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an ingestion surface for untrusted data, though it is considered safe due to the protocol's design.
Ingestion points: Solana blockchain data is fetched via createSolanaRpc and fetchCollection in SKILL.md.
Boundary markers: The skill does not explicitly define prompt boundary markers for fetched data, but the protocol avoids external URI/JSON metadata fetches, significantly reducing the attack surface.
Capability inventory: The skill includes capabilities to send and confirm transactions (sendAndConfirmTransactionFactory) as seen in SKILL.md.
Sanitization: The @gl-36/sdk performs structural validation and layout decoding on account bytes before they are processed by the agent.

gl36