freeride

FreeRide is an essential productivity and engineering utility for OpenClaw users aiming to leverage free-tier AI models while maintaining high availability. The skill automates the complex process of selecting, ranking, and configuring LLM providers available via OpenRouter. By creating a robust fallback hierarchy, FreeRide ensures that if a specific model hits rate limits, the system seamlessly transitions to the next best available free alternative without manual intervention.

• Automatically manages openclaw.json configurations to set primary models and fallback arrays.
• Includes a daemonized freeride-watcher to monitor rate-limit events and perform real-time model rotation.
• Provides a command-line interface to list available models, switch primary providers, and refresh cached configuration states.
• Supports custom fallback counts to balance performance and reliability for different agent workloads.
• Preserves critical environment variables, gateway settings, and custom instructions during configuration updates.

This tool is intended for developers and power users who want to reduce AI infrastructure costs without sacrificing the continuity of their agentic workflows. It requires an active OPENROUTER_API_KEY and expects the OpenClaw gateway to be restarted after configuration changes to apply settings. Typical inputs involve CLI commands such as 'freeride auto' to initialize the system or 'freeride switch' to pin a specific model. Outputs include updated configuration files and logs of successful model transitions. Users should be aware that the quality and availability of free models vary; the skill's auto-ranking logic is designed to mitigate this unpredictability by favoring established models like Qwen or other OpenRouter-verified free tiers. Always verify state changes using 'freeride status' or the internal /status command within the OpenClaw interface.