opensea-api

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). SKILL.md requires using OpenSea API/CLI/MCP to ingest API response fields (user-generated NFT/collection metadata) into the agent’s runtime LLM context via the response output path (stdout wrapped by --- BEGIN/END OPENSEA API RESPONSE ---), which is outsider-authored free text from OpenSea users.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill exposes explicit crypto/NFT transaction functionality: it can build mint and deploy transactions (get_mint_action / deploy_seadrop_contract) that return transaction data ("to, data, value") meant to be signed and submitted, and provides scripts to build asset transfer transactions. It also includes wallet-scoped write operations (e.g., write:orders → cancel_orders, write:drops, write:wallets) — all are specific blockchain/crypto actions (wallets, transactions, orders). Even though some endpoints return unsigned tx data, these are specific crypto transaction tools (not generic HTTP or browser tools), so this skill provides direct crypto financial execution capability.