COMandA is a Go-based inference engine designed for orchestrating complex agentic workflows by chaining together Large Language Model (LLM) operations. It empowers users to build sophisticated, multi-step processes using simple YAML configurations, enabling the combination of strengths from various LLM providers and custom actions.

How It Works

COMandA processes user-defined YAML "recipes" that specify sequences of operations. Each step in a recipe can involve inputs (files, URLs, screenshots, database queries), LLM calls (supporting multiple providers like OpenAI, Anthropic, Google, Ollama, X.AI), custom actions, and output handling. The engine intelligently chains these steps, passing outputs from one to the inputs of the next, and supports parallel execution of independent steps for performance gains. It also includes advanced features like web scraping, image analysis, and database integration.

Quick Start & Requirements

• Install: Download pre-built binaries from GitHub Releases or install via go install github.com/kris-hansen/comanda@latest.
• Configuration: Requires API keys for supported LLM providers, configured via a .env file or the comanda configure command.
• Resources: No specific hardware requirements mentioned beyond standard OS compatibility for binaries. Setup involves configuring API keys.
• Docs: Examples are available in the examples/ directory.

Highlighted Details

• Supports a wide array of LLM providers including local models via Ollama.
• Features robust web scraping capabilities and direct URL input support.
• Includes image analysis for common formats and screenshots, with automatic optimization.
• Offers a server mode for exposing workflows as an HTTP API with authentication and CORS configuration.
• Supports database integration (PostgreSQL) for input and output operations.
• Enables parallel processing of independent workflow steps.

Maintenance & Community

The project is actively maintained by Kris Hansen. Contributions are welcomed via pull requests.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

Currently, only PostgreSQL is supported for database operations. The roadmap indicates planned support for branching logic, routing, and additional providers, suggesting some advanced orchestration features are still under development.