accessing-data

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent to query the external MCP documentation server at runtime for up-to-date answers (thus allowing remote content to directly influence agent responses): https://sui.mcp.kapa.ai

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

• Direct money access detected (high risk: 1.00). The skill explicitly documents and exposes transaction-submission APIs (gRPC and GraphQL RPC), mentions the transaction_execution_service, and names client primitives (SuiGrpcClient / SuiGraphQLClient) plus GraphQL's executeTransaction / simulateTransaction. Those are specific blockchain transaction-sending interfaces (i.e., "send transaction") which enable moving on-chain assets — so this skill grants direct financial execution capability.