integrating-jupiter

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

• Third-party content exposure detected (high risk: 0.90). This skill's SKILL.md "Fresh Context Policy" explicitly requires the agent to fetch and treat external docs/specs from public sites (e.g., dev.jup.ag OpenAPI/docs and linked GitHub URLs) as the source of truth before executing playbooks, which means the agent will ingest untrusted, third‑party web content that can influence subsequent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill's "Fresh Context Policy" explicitly requires fetching live docs/specs at runtime (e.g., https://dev.jup.ag/openapi-spec/ultra/ultra.yaml or https://dev.jup.ag/docs/ultra/index.md) and treating that fetched content as the source of truth, so external URLs directly control the agent's instructions.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly a Jupiter crypto integration: it documents swap/quote endpoints (Ultra Swap GET /order -> POST /execute), order/limit/recurring endpoints (Trigger /createOrder + /execute, Recurring /createOrder + /execute), lending endpoints that return unsigned VersionedTransactions (/earn/deposit, /earn/withdraw), send endpoints (/craft-send, /craft-clawback), and other market/order endpoints (prediction markets /orders). The prompt even includes a signAndSend implementation (signing VersionedTransaction with a Keypair and sending to RPC). These are specific crypto transaction/asset operations (swaps, sends, order execution, signing), i.e. direct financial execution capability.