DEV Community

ddamno
ddamno

Posted on

InCoder | JetBrains Plugin

#langchain #ai #opensource #github

What is InCoder?

InCoder is a plugin designed for JetBrains IDEs — including IntelliJ IDEA, PyCharm, and others in the ecosystem. It seamlessly integrates advanced Large Language Models (LLMs) into your development workflow, offering in-editor code generation, comprehension, and editing assistance.

Whether you're using OpenAI, Claude, or a local model with Ollama, InCoder keeps everything inside your IDE — no external chat windows, no token quotas, no distractions.

demo chat

Why I Built InCoder

I spend most of my day inside IntelliJ Community Edition. I didn’t want to pay for Ultimate just to access AI features. And I didn’t want to rely on bloated, cloud-locked tools that felt disconnected from how I actually code.

What I really wanted was:

  • Full control over the LLM and its behavior
  • The ability to use local models (Ollama) or Claude via API
  • A workflow that stays in the IDE and understands my entire project

So I built InCoder.

What It Does

  • Connects to OpenAI, Anthropic (Claude), and Ollama (local models like LLaMA 3, Mistral, etc.)
  • Can be extended to support other LLM providers
  • Reads project files automatically to understand context
  • Lets you define and tweak system prompts
  • Gives LLMs access to tools for:
    • File reading/writing
    • Creating patches or merge requests
    • Running CLI commands (with user approval)

Who It's For

  • Developers using JetBrains Community Edition (or any JetBrains IDE)
  • Those who want full control over their AI assistant
  • People frustrated with Junie, Copilot, or plugin limitations
  • Tinkerers who want to experiment with local models or self-hosted APIs

My Setup

Right now, I use Claude Sonnet via Anthropic’s API — and it works beautifully. I’ve also tested InCoder with Ollama running local LLMs like LLaMA 3. The performance is solid, the integration is seamless.

But what really makes it powerful is flexibility.

  • I can run custom CLI commands from prompts (with confirmation)
  • I can scaffold components, refactor, or generate boilerplate code
  • I’ve built full React + TypeScript frontends this way — with the LLM guiding architecture, structure, and code

Final Thoughts

I didn’t set out to build a product — I built InCoder because I needed a tool that just worked the way I code. It's fast. Hackable. Private. And designed to integrate natively with the JetBrains environment.

If that sounds like your style, give it a try.

👉 Install InCoder
🛠️ Contribute or check out the code

Happy hacking!
— Damiano

Top comments (0)