Repolode
Submit
kris-hansen

kris-hansen/comanda

A terminal-native CLI that turns plain-English descriptions into YAML workflows orchestrating Claude Code, Codex, Gemini CLI and other coding agents with quality gates.

View on GitHubWebsite
Stars
332
Forks
27
Open issues
2
Last commit
5 days ago

What is it?

Comanda is a Go command-line runtime for running AI coding agents as repeatable, reviewable workflows inside a repository. You describe an outcome in English, Comanda generates a YAML program, and you can render that program as a Mermaid or terminal graph, run it, and refine it with further plain-English feedback.

Workflows can coordinate Claude Code, Gemini CLI, OpenAI Codex, Kimi Code, API models and local models, passing work between steps through files or standard input and output. Agentic loops persist state, support checkpoints and resume, and run user-defined quality gates such as tests, syntax checks and security checks after each iteration, with retry, abort or skip policies.

The project also covers repository context: Git worktrees for parallel implementations, codebase indexes with diff and update support, a local knowledge graph stored in a SQLite FTS5 memory database, and an MCP server that exposes workflows as tools. Ruby and Rails repositories are indexed automatically, and PostgreSQL DDL is parsed into schema graph nodes.

Key features

  • 01Generate YAML workflows from plain-English descriptions and modify them with natural-language feedback.
  • 02Render workflows as terminal or Mermaid graphs for inspection.
  • 03Run agentic loops with checkpoints, resume, and configurable quality gates.
  • 04Coordinate Claude Code, Gemini CLI, OpenAI Codex, Kimi Code and API or local models in one workflow.
  • 05Build codebase indexes and a local knowledge graph, with incremental updates.
  • 06Expose workflows as MCP tools and run them in shell pipelines.

Use cases

  • Having a coding agent implement a feature repeatedly until your tests and security checks pass.
  • Running several coding agents in parallel on a design or diff and synthesizing their reviews.
  • Feeding git diff output into a multi-agent review workflow in CI or a terminal pipeline.
  • Giving agents durable project context from a codebase index and stored architecture decisions.

Quick start

  1. 1

    Install with Homebrew (macOS)

    brew install kris-hansen/comanda/comanda
  2. 2

    Or install with Go

    go install github.com/kris-hansen/comanda@latest
  3. 3

    Generate a workflow from a plain-English description

    comanda generate feature-loop.yaml "Implement this feature until tests and security checks pass"
  4. 4

    Run the workflow in your repository

    comanda process feature-loop.yaml

Things to know

  • Several workflow steps depend on external coding agents (Claude Code, Codex, Gemini CLI) or API keys that you must already have set up.

In collections

Open-source AI agent frameworks & appsAI coding assistants & agentsMCP serversCLI & terminal tools

Topics

#ai-agents#cli#workflow-automation#llm#orchestration#yaml#mcp#code-review#agents#ai

FAQ

What is comanda?
A terminal-native CLI that turns plain-English descriptions into YAML workflows orchestrating Claude Code, Codex, Gemini CLI and other coding agents with quality gates. Comanda is a Go command-line runtime for running AI coding agents as repeatable, reviewable workflows inside a repository. You describe an outcome in English, Comanda generates a YAML program, and you can render that program as a Mermaid or terminal graph, run it, and refine it with further plain-English feedback.
Is comanda free to use? What license does it have?
comanda is open source under the MIT license. Check the license terms before using it in commercial products.
Is comanda still maintained?
Actively maintained. The last commit to comanda was on Nov 2, 2026.
What can I use comanda for?
Having a coding agent implement a feature repeatedly until your tests and security checks pass. Running several coding agents in parallel on a design or diff and synthesizing their reviews. Feeding git diff output into a multi-agent review workflow in CI or a terminal pipeline. Giving agents durable project context from a codebase index and stored architecture decisions.
How do I install comanda?
Install with Homebrew (macOS): brew install kris-hansen/comanda/comanda Or install with Go: go install github.com/kris-hansen/comanda@latest Generate a workflow from a plain-English description: comanda generate feature-loop.yaml "Implement this feature until tests and security checks pass" Run the workflow in your repository: comanda process feature-loop.yaml
Where can I get comanda?
The source code and installation instructions are on GitHub: https://github.com/kris-hansen/comanda

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects