DEV Community

Stephan Coertzen
Stephan Coertzen

Posted on

Introducing spec-kit: Spec-Driven Development + Ollama & OpenRouter Integration

Hey folks,

I’ve been working on some enhancements to spec-kit
, and I’m excited to share them with you. These updates bring in CLI, Ollama, and OpenRouter support to supercharge spec-driven dev workflows.

πŸ” What is spec-kit?

Spec-kit is a toolkit for Spec-Driven Development, helping developers go from idea β†’ spec β†’ implementation with better alignment and iterative clarity.

It supports:

Multiple AI coding agents (Claude Code, Gemini CLI, GitHub Copilot, Cursor, etc.)

Cross-stack / tech independence: not tied to one language or framework

Lightweight requirements (Python 3.11+, Git, Linux/macOS or WSL2 on Windows, and an AI agent)

✨ What I’ve added

With my contributions, spec-kit now also allows:

CLI support: run spec-kit directly from the terminal

Ollama integration: use local LLMs as backend agents

OpenRouter support: connect to multiple hosted models with one interface

This means you can now choose fully local, fully cloud, or hybrid workflows.

πŸš€ Core CLI Commands

Here are some of the most useful commands to try out:

  1. Specify your project
speckit specify
Enter fullscreen mode Exit fullscreen mode

Use AI to flesh out your requirements and specs. This is where ideas become structured, testable specifications.

  1. Generate a plan
speckit plan
Enter fullscreen mode Exit fullscreen mode

Automatically generate a high-level plan of implementation steps from your specs.

  1. Break down into tasks
speckit tasks
Enter fullscreen mode Exit fullscreen mode

Turn your plan into actionable development tasks β€” perfect for managing iterations and team work.

⚑ Example with Ollama & OpenRouter

Run a spec using a local Ollama model

speckit specify "Write a todo list application" --ai ollama
Enter fullscreen mode Exit fullscreen mode

Or use OpenRouter to run via a hosted model

speckit plan "Tech Stack: Docker, MongoDB, React, Vite, NodeJS" --ai openrouter
Enter fullscreen mode Exit fullscreen mode

πŸ’‘ Why this matters

Faster prototyping β†’ from raw idea to structured plan in minutes

Flexible model choice β†’ local (Ollama) or hosted (OpenRouter)

Better workflows β†’ specs drive planning and tasking, not just ad-hoc coding

πŸ‘€ What’s next

I’d love feedback on:

Other LLM providers to support

How you’re using Ollama / OpenRouter in practice

Improvements for the CLI flow (esp. specify β†’ plan β†’ tasks pipeline)

Contributions and ideas are very welcome πŸ™Œ

πŸ‘‰ Try it here: https://github.com/jfscoertzen/spec-kit#

Top comments (0)