The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the consumption of untrusted data from the TON blockchain.
Ingestion points: get_nfts and get_nft tools in SKILL.md retrieve external NFT metadata from the blockchain.
Boundary markers: The instructions lack specific delimiters or system instructions to ignore potential commands embedded in NFT metadata.
Capability inventory: The skill utilizes send_raw_transaction (documented in SKILL.md), allowing it to perform state-changing operations on the blockchain.
Sanitization: No data sanitization or validation logic is defined for the retrieved NFT metadata before it is processed by the agent.

ton-nfts