comanda

What This Skill Does

Comanda is a powerful declarative workflow engine that enables OpenClaw users to orchestrate complex AI operations using YAML-based pipelines. It abstracts away the intricacies of multi-model chaining, parallel processing, and data piping, allowing you to treat AI workflows as modular code. Whether you are automating a code review process, summarizing long-form documents through multi-step refinement, or building self-modifying meta-workflows, Comanda provides a structured CLI environment to define, visualize, and execute these tasks. It natively supports a broad ecosystem of LLM backends including OpenAI, Anthropic, Google, and local models via Ollama.

Installation

To integrate this skill into your environment, use the OpenClaw management CLI:

1. Ensure you have the OpenClaw agent installed.
2. Run: clawhub install openclaw/skills/skills/kris-hansen/comanda
3. After installation, execute comanda configure to securely set up your necessary API keys for the model providers you intend to use.

Use Cases

Comanda is designed for users needing repeatable, robust AI automation. Typical use cases include:

• Multi-Model Orchestration: Running parallel analysis steps where one model checks for security vulnerabilities while another evaluates performance.
• Automated Research Pipelines: Chaining a document scraper, an extraction step, and a final summarization step into a single YAML configuration.
• Agentic Code Development: Leveraging meta-workflows where the AI agent creates its own workflow file based on a natural language requirement and subsequently executes it.
• Workflow Documentation: Using the chart functionality to visually audit how data flows through your AI pipeline.

Example Prompts

1. "Use Comanda to create a workflow in code-review.yaml that pulls from my current directory, analyzes it for security risks with Claude, and writes the results to a file named security-audit.txt."
2. "Show me the structure of my current summary.yaml workflow and verify if it is configured correctly for execution."
3. "Execute the workflow defined in research-pipeline.yaml using the latest GPT-4o model and display the final output."

Tips & Limitations

• Version Control: Treat your YAML workflow files as code. Commit them to version control to maintain a history of your AI automation logic.
• Efficiency: Use the parallel-process feature to significantly reduce latency when performing multiple independent evaluations on the same input data.
• Limitations: Comanda requires correctly configured environment variables or API keys. It currently relies on file-system access to read/write workflows, so ensure your agent has appropriate permissions to the target directory. Avoid using highly sensitive raw credentials directly in YAML files; use environment variables wherever possible.