@6522/tangcore-agent

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow explicitly treats external research surfaces (e.g., Perplexity/ChatGPT/news-aggregator output and pitch decks) as unverified leads in its pipeline intake/classification, meaning outsider-authored free text can enter the evidence ledger for re-verification and influence downstream reasoning.

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

• Direct money access detected (high risk: 1.00). The skill explicitly lists trading/exchange data and trading tools in its Tool Priority Map (e.g., "perp depth/trading → hyperliquid" and exchange/on-chain tooling like okx), which are specific financial/trading APIs rather than generic data callers. Those entries indicate integration with platforms whose primary purpose is trading/market execution, so the skill includes explicit financial execution capability risk.