fabro-workflow-factory

What This Skill Does

The fabro-workflow-factory skill empowers your OpenClaw agent to orchestrate complex coding pipelines using Fabro, a Rust-based workflow engine. It treats AI development not just as a single prompt, but as a structured, repeatable process. By utilizing Graphviz DOT files, you can map out sophisticated agent logic that includes branching, loops, and critical human-in-the-loop approval stages. The skill provides direct control over model selection, allowing you to define distinct model routing rules via CSS-like stylesheets for different parts of your workflow, ensuring cost and performance optimization throughout the execution lifecycle.

Installation

To install this skill, use the following command in your terminal:

clawhub install openclaw/skills/skills/adisinghstudent/fabro-workflow-factory

Ensure you have the Fabro binary installed on your system by running the installation script provided by the project: curl -fsSL https://fabro.sh/install.sh | bash. Once installed, initialize your project using fabro init to configure the necessary environment for workflow execution.

Use Cases

• Automated Refactoring: Map a series of nodes that perform static analysis, code transformation, and final unit testing, requiring a human 'approve' gate before merging changes.
• Multi-Agent Systems: Deploy a workflow where an architect node designs the schema, a developer node writes the implementation, and a QA node creates the test suite.
• CI/CD Pipelines: Build custom deployment workflows that trigger cloud sandboxes to preview applications, exposing ports for verification before moving to production.
• Model Comparison: Route different branches of a task through different LLM models (e.g., Claude 3.5 Sonnet vs. Haiku) to compare output quality or cost-efficiency within the same automated graph.

Example Prompts

1. "Initialize a new Fabro workflow in the current directory and create a DOT file for a simple code review loop that asks for my approval before committing changes."
2. "Run the feature-build.dot workflow and stream the output to the terminal so I can monitor the multi-model execution steps."
3. "Show me the retrospective for run ID 8492 to check the total cost and duration of the last automated test session."

Tips & Limitations

• Tip: Use fabro ssh <run-id> to inspect the environment if a workflow fails at a specific node, allowing for deep debugging of the sandbox state.
• Tip: Organize your DOT files into separate modules for complex pipelines to keep your graph logic readable and maintainable.
• Limitation: This skill requires a persistent connection to the Fabro service; ensure your environment allows outbound network traffic to the required APIs.
• Limitation: Large graphs can become complex; always test individual nodes with smaller, isolated DOT files before deploying full-scale, multi-stage agent workflows.