javascript-sdk

What This Skill Does

The javascript-sdk skill enables OpenClaw agents to interface directly with the inference.sh platform using a powerful TypeScript-first library. It acts as a bridge between your AI agent and a massive ecosystem of over 150+ AI models. Whether you are building complex agentic workflows, generating media, or running advanced inference tasks, this skill simplifies the process by handling authentication, API communication, file uploads, and streaming updates. It supports sophisticated features like fire-and-forget execution for long-running tasks, real-time log streaming for observability, and automated file handling to ensure seamless data pipelines between your local filesystem and the cloud.

Installation

To integrate this skill into your workflow, use the following terminal command: clawhub install openclaw/skills/skills/okaris/javascript-sdk

Once installed, ensure you have your inference.sh API key ready. You can configure the client by either passing the key directly to the constructor or by setting it in your environment variables for better security. For developers building frontend-facing applications, the SDK provides built-in support for proxy-based requests, ensuring your credentials remain hidden.

Use Cases

This skill is highly versatile and serves several primary use cases:

• Agent Integration: Build custom AI agents that can trigger model inferences on-demand as part of a larger plan.
• Content Generation Pipelines: Automate image or text generation workflows where the agent needs to process inputs and handle model outputs programmatically.
• Application Development: Quickly prototype and ship React or Next.js applications that require direct access to high-performance AI models.
• File-Driven AI Apps: Utilize the SDK's built-in file handling to automatically upload assets from the local environment to the inference engine, process them, and return the result.

Example Prompts

1. "Initialize a new inference client using my API key and trigger the flux-schnell model with a prompt about a futuristic cityscape."
2. "Run the image-processor app on the file located at ./assets/input.jpg and stream the execution logs back to me in real-time."
3. "Set up a task to execute the veo-3-1-fast model as a fire-and-forget request, then let me know the task ID once the process has started."

Tips & Limitations

• Security: Avoid hardcoding your API keys directly in source code. Always use environment variables for sensitive credentials.
• Versioning: When running apps, specify the model version if needed, or use the default namespace/name format for standard deployments.
• Infrastructure: Be aware that certain models may require specific infra configurations; check the app documentation on inference.sh for requirements.
• Environment: While Node.js 18.0.0+ is recommended for the best experience, ensure your network allows for the outbound HTTPS requests required by the inference API.