add-tool

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

• Potentially malicious external URL detected (high risk: 1.00). The skill makes runtime GET requests to https://agenticstack.sh/api/json/categories (and related endpoints https://agenticstack.sh/api/json/tools and https://agenticstack.sh/api/json/categories/{category}), and the returned category list and feature_definitions are directly used to determine the prompt structure and the exact agent_features keys, so the external content controls the agent's instructions at runtime.