hawk-strategy

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

• Third-party content exposure detected (high risk: 0.72). The required runtime workflow uses an LLM “decision_prompt” that includes {{signal_hawk_signals}} from the hawk_signals external scanner (via `context:
• type: signal), and that scanner’s data/reasons/trend4h/...fields originate from outsider-authored market/leaderboard API content fetched at runtime, which can contain free-form text (e.g.,reasons`) that the LLM ingests into its context.

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

• Potentially malicious external URL detected (high risk: 0.90). The README's install step instructs using curl to fetch Python skill files from https://raw.githubusercontent.com/Senpi-ai/senpi-skills/main/hawk/$f and then running the downloaded script (nohup python3 .../hawk-producer.py), which clearly downloads remote code that will be executed as part of setup/runtime.

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 for crypto assets (BTC, ETH, SOL) whose primary purpose is to enter and manage positions: it states "Producer enters. DSL exits." and defines entry gates, scoring,phase exits, ladders, and timeouts for real trade execution. Although it mentions data calls (market_get_asset_data, leaderboard_get_markets), the skill is specifically designed for market trading (buy/short) rather than a generic data or automation tool. Under the policy, this is a market-order trading capability (direct financial execution authority).