Security audit
Security checks for vulnerabilities and agentic risk
This skill is clearly for prediction markets, but it teaches agents to make real HBAR/USDC bets without clear confirmation, limits, or risk controls.
Review this carefully before installing. Only use it when you intend an agent to interact with real prediction-market services, and require explicit approval for every bet, market creation, claim, or credential use. Set small spending limits, verify market details and balances yourself, and treat API keys/JWTs as sensitive secrets.
The skill explicitly instructs agents to place bets, create markets, and use HBAR/USDC without any warning, confirmation requirement, spending limits, or notice that transactions may be irreversible and financially risky. In an agent-skill context, this increases the chance of autonomous fund movement or market participation without informed user consent.
The skill directs the agent to transmit data to external services and obtain credentials/JWTs from third-party endpoints, creating a path for data exfiltration, account creation, and subsequent authenticated financial actions. In this context, the transmission is more sensitive because it is directly chained to betting and market operations involving real assets.
curl -X POST https://onlyflies.buzz/clawswarm/api/v1/agents/register \
-H "Content-Type: application/json" \
-d '{"name": "YourAgent", "capabilities": ["predictions", "analysis"]}'
No suspicious patterns detected.