poe-chat

What This Skill Does

The Poe Chat skill allows OpenClaw users to interface with a wide variety of state-of-the-art AI models—including Gemini, GPT, Claude, Kimi, and Deepseek—directly from their workspace. By leveraging the Poe API, this skill acts as a dynamic gateway that intelligently maps natural language triggers to specific model IDs. It handles the complexity of model selection, file uploads, and response formatting, ensuring users get the best possible output for their specific task without leaving the OpenClaw environment.

Installation

To integrate this skill, ensure you have the OpenClaw framework installed. Run the following command in your terminal: clawhub install openclaw/skills/skills/longmans/poe-chat

Once installed, you must install the required dependencies found in scripts/requirements.txt via pip install -r scripts/requirements.txt. Finally, secure your authentication by setting the POE_API_KEY as an environment variable or passing it directly during execution.

Use Cases

• Research and Summarization: Upload multi-page PDFs or technical documentation and ask specific questions using @claude-3-opus or similar models.
• Programming Assistance: Quickly generate code snippets or debug logs by targeting specialized coding models like @deepseek-coder.
• General Knowledge: Use @gpt-4o or @gemini for broad inquiries, creative writing, or drafting emails.
• Automated Routing: Let the skill automatically route your complex queries to the most cost-effective and capable model available based on your internal configuration.

Example Prompts

1. "@gemini 请阅读并总结我上传的这份财务报告的核心风险点。" (Invokes Gemini with a file upload)
2. "@gpt 写一段关于Python异步处理的代码示例，需要包含错误处理逻辑。" (Invokes the default GPT model)
3. "@claude-opus-4.6 分析这段日志，指出为什么这个程序出现了内存溢出。" (Invokes a specific high-end Claude model)

Tips & Limitations

• Model Selection: Always check the auto-generated models.json file to understand the current available IDs. The skill follows a priority logic: it favors "Flash" or base models over "Pro/Ultra" models to save resources unless you explicitly request a high-end version.
• Caching: The skill caches the model list for 1 hour; if you notice new models appearing on Poe but not in your list, delete the local models.json file to force a refresh.
• Privacy: Be mindful that any file uploaded via the --file flag is sent to the Poe service provider for processing; ensure sensitive internal data is redacted before uploading.