helixa

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

• Third-party content exposure detected (medium risk: 0.30). The workflow queries agent directories and search results from api.helixa.xyz, which contain user-generated agent profiles and metadata (such as names, narratives, and traits) that are consumed by the LLM at runtime without first selecting a specific item.

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

• Direct money access detected (high risk: 1.00). The skill documentation includes explicit capabilities for executing blockchain transactions and managing micropayments, specifically handling $1 USDC payments via x402 for minting agent identities and executing direct contract mint transactions using Foundry's cast send with ETH value on Base mainnet.