The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection surface identified through the skill's reliance on external market signals to trigger automated financial transactions.
Ingestion points: scripts/emerging-movers.py and scripts/sm-flip-check.py ingest market concentration data and trader sentiment from Hyperliquid via the leaderboard_get_markets MCP tool.
Boundary markers: The skill does not implement delimiters or secondary validation to distinguish between legitimate market trends and artificial data manipulation (e.g., wash trading or metadata spoofing) intended to trigger the bot's 'FIRST_JUMP' entry logic.
Capability inventory: The system possesses significant financial capabilities, including opening and closing leveraged positions via scripts/open-position.py and the create_position MCP tool.
Sanitization: While the skill applies quantitative filters (such as rank history and contribution velocity), it lacks security-focused sanitization to prevent the automated execution of malicious trades triggered by data poisoning.

wolf-strategy