A distributed inference network that allows users to share idle Ollama capacity or consume community compute when local resources or API credits are unavailable.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install idleclaw
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install idleclaw using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is IdleClaw?
IdleClaw is a decentralized compute sharing layer built specifically for the Ollama ecosystem. It functions by connecting distributed nodes—users with idle GPU or CPU capacity—to a central routing server that manages inference requests from the community. By incorporating this into your Openclaw Skills, you can transform your local machine into a contributor node or leverage external community hardware to run large language models without relying on expensive centralized API providers.
The system is designed with a focus on privacy and security, acting as a relay rather than an execution environment. This ensures that while compute is shared, your local filesystem and sensitive data remains protected. It is an ideal solution for developers looking to maximize hardware utility and maintain access to LLMs even when their primary credits are exhausted.
IdleClaw Use Cases
- Sharing idle GPU or CPU resources with the developer community when your workstation is not in use.
- Accessing specific LLM models that you haven't downloaded locally by utilizing community nodes.
- Offloading inference tasks to the network when your local machine is under heavy computational load.
- Maintaining workflow continuity by using community compute as a fallback when API credits run out.
How IdleClaw Works
- The contributor starts a local node which establishes a WebSocket connection to the IdleClaw routing server.
- The node registers its list of available models pulled from the local Ollama instance.
- A consumer sends an inference request (prompt and model name) to the routing server via an HTTP POST request.
- The routing server identifies an available contributor node that hosts the requested model and forwards the JSON inference parameters.
- The contributor's node relays the request to their local Ollama instance, receives the generated tokens, and streams them back through the routing server to the consumer.
IdleClaw Setup
To get started, ensure you have Python 3 and Ollama installed on your system. Run the installation script to set up the necessary dependencies for this part of your Openclaw Skills:
cd "$SKILL_DIR" && bash install.sh
To share your idle capacity with the network:
python scripts/contribute.py
To use community inference for a prompt:
python scripts/consume.py --model <model-name> --prompt "<your message>"
IdleClaw Data Schema & Taxonomy
IdleClaw uses a transient data model where no user data is persisted beyond the active session. All communication is handled via JSON text over secure channels.
| Data Type | Description | Persistence |
|---|---|---|
| Inference Params | Model name, messages, tools, and options | Session Only |
| Node Metadata | Node ID and available model list | In-Memory |
| Responses | Role, content, and tool calls | Streamed, Not Stored |
| Validation | Whitelisted keys (model, messages, stream, etc.) | Real-time Only |
IdleClaw Advanced Features
- Real-time streaming support using Server-Sent Events (SSE) for low-latency token delivery.
- Support for custom routing servers, allowing teams to set up private, self-hosted inference clusters.
- Comprehensive input sanitization that strips unknown keys and enforces message length limits.
- Multi-node support per IP with integrated rate limiting to prevent network abuse.
- Server-side tool execution safeguards including schema validation and execution timeouts.
SKILL.md
Loading
Related Openclaw Skills
Polymarket User Analyzer
Analyze any Polymarket trader's wallet and strategy to uncover winning patterns and performance metrics using Openclaw Skills.
ClawPrint Verify
A reverse-CAPTCHA mechanism designed to verify that a conversation partner is an AI agent rather than a human through high-speed computation and pattern recognition.
ClawPrint Verify
A reverse-CAPTCHA security tool that issues high-speed computational challenges to verify if a user or agent is a real AI rather than a human.
ClawRent Web Scraping
Monetize your computer's idle hardware by performing automated web scraping tasks to earn passive income.
SafeFlow Sui Skill
A specialized AI agent skill for managing SafeFlow shared contracts on the Sui blockchain with secure owner-assisted provisioning.
Nano Banana OpenRouter Image Generation
A specialized image generation skill leveraging Google's Gemini 2.5 Flash Image models through the OpenRouter API.
METADATA
Featured*
