clawflows

What This Skill Does

ClawFlows is the orchestration engine for the OpenClaw ecosystem, designed to bridge the gap between individual, isolated skills and complex, multi-step automations. Instead of manually invoking one skill after another, ClawFlows allows you to define portable YAML-based workflows. It operates on an abstraction layer called 'capabilities.' By referencing capabilities rather than specific skill names, ClawFlows ensures your automations remain functional regardless of which specific implementation of a service you have installed. It handles the data flow between steps, conditional logic, and scheduling, effectively turning your OpenClaw agent into a powerful autonomous pipeline manager.

Installation

To begin using ClawFlows, ensure you have Node.js installed on your system. You can install the CLI tool globally by executing the following command in your terminal:

npm i -g clawflows

Once installed, you can verify your environment by running clawflows --help. To integrate existing skills, ensure they have a CAPABILITY.md file defined, which informs the ClawFlows engine how to map your skill methods to the automation pipeline.

Use Cases

ClawFlows shines in complex data processing tasks. A classic use case is a 'Morning Brief' automation: it can pull news from a social-search capability, filter the data, summarize it using an LLM skill, and send the final report via an email capability. Another common use case is competitive tracking, where an automation fetches the latest video performance from the youtube-data capability, updates a local database using the database capability, and then triggers a chart-generation step to visualize the growth trends over time. It is perfect for users who want to move beyond single-purpose commands into reliable, repeatable background tasks.

Example Prompts

1. "ClawFlows, search for any available automations that can help me track competitor growth on YouTube and visualize it in a chart."
2. "Please install the 'morning-brief' automation and verify that I have all the necessary skill capabilities installed on my system."
3. "Run the 'youtube-competitor-tracker' automation in dry-run mode to see which data would be fetched before I perform the full execution."

Tips & Limitations

When building automations, always use the --dry-run flag to validate data flow and step logic before letting a workflow interact with external APIs or databases. Keep in mind that ClawFlows relies on the quality of the CAPABILITY.md definitions provided by skill authors. If a required capability is missing or misconfigured, the automation will fail at the installation or runtime check phase. For advanced users, remember that variables are passed between steps using the capture key, allowing you to use the output of one capability as the input for the next, creating a truly reactive and dynamic system.