swarm-coding-skill

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

• Potentially malicious external URL detected (high risk: 1.00). The orchestrator performs runtime POSTs to OpenRouter at https://openrouter.ai/api/v1/chat/completions to fetch LLM responses that directly determine the planner/worker prompts and produced code (and the skill requires an OPENROUTER_API_KEY), so this external endpoint controls agent instructions and generated code.

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

• Direct money access detected (high risk: 1.00). The skill explicitly includes crypto/web3 wallet infrastructure and smart-contract generation. The "Privy Integration" section says the swarm automatically includes Privy authentication and wallet infrastructure (frontend use of @privy-io/react-auth, backend /auth/callback, and a simulated fallback), and example prompts include "Generate a smart contract that holds ClawCredit limits and allows x402 payments." These are specific crypto/blockchain tools/features (wallets and smart-contract payment logic), which meet the criteria for Direct Financial Execution capability. Although the skill is primarily a general code generator, it explicitly provisions wallet/auth and can generate smart-contract/payment code, so it should be flagged.