idleclaw

What This Skill Does

IdleClaw transforms your local Ollama instance into a participant in a distributed inference network. It allows you to toggle between two distinct modes: contributing and consuming. When in contribute mode, your idle local GPU/CPU resources serve inference requests from the community, effectively decentralizing AI compute. When in consume mode, if you find your personal API credits are exhausted, the skill allows you to route your inference tasks to the community network, ensuring that your AI workflows remain uninterrupted.

Installation

To integrate IdleClaw into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/futurejunk/idleclaw

Ensure that you have a functioning instance of Ollama running on your local machine with at least one model pulled before attempting to launch the scripts. The skill relies on standard Python 3 environments and requires access to your local Ollama API endpoints.

Use Cases

1. Community Contribution: Users with powerful hardware or idle servers can contribute to the democratized AI movement by serving models to others.
2. Backup Compute: Developers or researchers who hit rate limits or credit caps on proprietary LLM providers can seamlessly failover to the community network.
3. Network Transparency: Administrators can monitor the health and model availability of the distributed network to understand the current state of decentralized AI inference.

Example Prompts

1. "OpenClaw, initiate contribution mode using IdleClaw so my idle GPU can help the community run Llama 3 models."
2. "I'm out of API credits. Please switch to consume mode and ask the community network to summarize this document for me: [Paste text here]."
3. "Run a health check on the network to see if there are any nodes currently hosting Mistral models."

Tips & Limitations

• Security: Since this skill connects to an external routing server, ensure your local Ollama instance is configured to only allow necessary traffic.
• Latency: Because this is a distributed network, response times may fluctuate based on the host node's hardware and network speed.
• Compatibility: Ensure your local Ollama models are up to date to maintain compatibility with the community-hosted versions.
• Stability: As a community-driven project, service availability is not guaranteed. It is best used as a secondary resource alongside your primary AI pipeline.