comanda

Declarative AI pipelines for the command line.

Define LLM workflows in YAML. Run Claude Code, Codex, and Gemini CLI in parallel. Version control everything.

🌐 comanda.sh — Getting started, features, and templates

Install

brew install kris-hansen/comanda/comanda

Or: go install github.com/kris-hansen/comanda@latest · Releases

Quick Start

comanda configure                              # Set up API keys
comanda generate "review this code for bugs"   # Generate workflow from English
comanda process workflow.yaml                  # Run a workflow

Example

parallel-process:
  claude:
    input: STDIN
    model: claude-code
    action: "Analyze architecture"
    output: $CLAUDE

  gemini:
    input: STDIN
    model: gemini-cli
    action: "Identify patterns"
    output: $GEMINI

synthesize:
  input: "Claude: $CLAUDE\nGemini: $GEMINI"
  model: claude-code
  action: "Combine into recommendations"
  output: STDOUT

cat main.go | comanda process multi-agent.yaml

Features

• Multi-agent — Claude Code, Gemini CLI, OpenAI Codex in parallel
• Agentic loops — Iterative refinement with tool use
• Codebase indexing — Persistent code context across workflows
• Git worktrees — Parallel execution in isolated branches
• All the I/O — Files, URLs, databases, images, chunking

See comanda.sh/features for full details.

Documentation

• Examples
• Multi-Agent Patterns
• Agentic Loops
• Server API

License

MIT