Data Generator / 数据生成器 - Generate training data from user instructions. Input: tool name + command list. Output: JSONL. / 根据用户指令生成训练数据。

This skill's code is small and readable, but there are important mismatches to be aware of: (1) SKILL.md advertises Excel input and a prompt_template parameter that the shipped Python file does not implement — so the documentation is unreliable. (2) The code reads MINIMAX_API_KEY / OPENAI_API_KEY and an API_URL env var (defaulting to http://127.0.0.1:8766). While the current code does not include the API key in request headers, it does send prompts and generated text to the configured API_URL. Before installing or using: - Review and run the Python file locally to confirm behavior. - Ensure API_URL (if set) points to a trusted host (prefer localhost or an internal LLM) and do not set secrets in environment unless you trust the endpoint. - Do not supply sensitive or proprietary data until you're satisfied where prompts are sent. - If you need Excel support or prompt customization, note the current implementation lacks those features — expect to modify or extend the code. If you want a fully aligned/safer experience, ask the author to fix the documentation or update the code so declared inputs match implemented behavior.

Type: OpenClaw Skill Name: data-generator-waai Version: 1.1.0 The skill is a specialized utility for generating synthetic AI training data for smart home device interactions (e.g., 'GreeQA', 'dev_control'). The code in `data_generator.py` constructs prompts and sends them to a configurable LLM endpoint (defaulting to localhost). While there is a minor discrepancy between the documentation in `SKILL.md` (which mentions Excel support) and the actual implementation (which lacks it), the code contains no evidence of malicious intent, data exfiltration, or unauthorized execution. It handles API keys from environment variables but does not appear to transmit them to external or suspicious domains.