egret-strategy

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

• Potentially malicious external URL detected (high risk: 0.90). The README instructs fetching and then executing remote code from the Senpi GitHub (e.g. "npx skills add https://github.com/Senpi-ai/senpi-skills" and curl "https://raw.githubusercontent.com/Senpi-ai/senpi-skills/main/egret/") which would download external code and run the producer daemon—a runtime/install-time fetch that results in executing remote code, so these URLs are high-risk.

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

• Direct money access detected (high risk: 1.00). The skill is explicitly a trading strategy intended to open and manage market positions: it specifies entry/exit behavior (maker-only entry / ensure_execution_as_taker, Phase 1/Phase 2 ladders, T0..T4 locks), risk/time-cut DSL for exits, and refers to market-related MCP calls and market data. Although it doesn't spell out a particular HTTP endpoint name, the document is specifically designed to place and manage market orders (buy/sell positions) and thus grants direct financial execution capability.