monad-swarm-agent

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

• Third-party content exposure detected (high risk: 1.00). The skill's Sentiment Agent explicitly states it monitors Twitter, Discord, and Telegram (and the README/config mentions TWITTER_BEARER_TOKEN and using Twitter/Discord APIs), meaning the agent is intended to fetch and interpret untrusted, user-generated social media content as part of its workflow.

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

• Direct money access detected (high risk: 1.00). The skill explicitly includes on-chain integration with Monad including configuration for MONAD_RPC_URL and MONAD_PRIVATE_KEY ("# For signing decisions") and examples of logging decisions with a decision hash/timestamp. That indicates explicit capability to sign and broadcast transactions on a blockchain (crypto signing). Under the rule set, crypto/blockchain signing constitutes Direct Financial Execution authority even if primary purpose is signaling/logging.