DOCSEARCHbyAlgolia
Submit Docs

DevoxxGenie IntelliJ IDEA Plugin

repository·master·Indexed 20 days ago

https://github.com/devoxx/devoxxgenieideaplugin

An agentic LLM code assistant for IntelliJ IDEA supporting local and cloud providers. It features Spec Driven Development (SDD) for autonomous task implementation, security scanning via Gitleaks, OpenGrep, and Trivy, and integration with the Model Context Protocol (MCP). The plugin provides RAG-based code search, inline completion, and support for the Agent Communication Protocol (ACP) to interact with external agents like Claude Code and GitHub Copilot.

Tokens
94.1K
Snippets
142
Records
499
Agent score
63%

What's inside DevoxxGenie

  1. Introduction to DevoxxGenie

    master
    DevoxxGenie is a Java-based LLM Code Assistant plugin for IntelliJ IDEA. It allows developers to integrate both local LLM providers (for privacy) and cloud-based LLM services (for power) directly into their IDE workflow. It is built using the Langchain4J library and is designed to enhance code quality, problem-solving, and learning through AI-driven features like code generation, agentic exploration, and spec-driven development.

  2. Overview of DevoxxGenie features

    master

    DevoxxGenie provides several advanced features designed for deep IDE integration and developer productivity:

    Context and Knowledge

    • MCP Support: Uses the Model Context Protocol to access external tools and services.
    • Project Scanner: Allows you to add your source code to the prompt context so the LLM has better awareness of your project.
    • Chat Memory: Maintains conversation history to ensure coherent multi-turn assistance.
    • Web Search: Augments LLM knowledge by retrieving real-time web search results.

    Developer Workflow Tools

    • Git Diff/Merge: Review and accept AI-generated code changes using a standard Git diff interface.
    • Token Cost Calculator: Helps manage and monitor API usage costs for cloud providers.

  3. Overview of DevoxxGenie

    master

    DevoxxGenie is a Java-based LLM Code Assistant plugin for IntelliJ IDEA. It supports a wide range of LLM providers, including local providers (Ollama, LMStudio, GPT4All, Llama.cpp, Nativ, Exo) and cloud-based providers (OpenAI, Anthropic, Mistral, Groq, Gemini, DeepInfra, DeepSeek, Kimi, GLM, OpenRouter, Cloudflare AI Gateway, Azure OpenAI, Amazon Bedrock, and NVIDIA).

    Key capabilities include:

    • Agentic Programming: Autonomous task implementation via Spec Driven Development (SDD).
    • Security Scanning: Integrated running of Gitleaks, OpenGrep, and Trivy.
    • Inline Completion: AI-powered code suggestions using Fill-in-the-Middle (FIM) models.
    • Agent Communication: Support for the Agent Communication Protocol (ACP) and CLI Runners to interact with external agents like Claude Code or GitHub Copilot.
    • RAG & Search: RAG-based context from vectorized project files and LLM-driven web search (Google, Tavily).

  4. Compare DevoxxGenie with other AI coding tools

    master
    DevoxxGenie is a free, open-source LLM code assistant plugin for IntelliJ IDEA that uses a BYOK (Bring Your Own Keys) model. Unlike subscription-based tools like GitHub Copilot or Cursor, DevoxxGenie allows you to connect to any LLM provider (cloud or local) using your own API keys, offering more flexibility in model selection and privacy control.

  6. Key Features of DevoxxGenie

    master

    DevoxxGenie offers a wide range of AI-assisted development features:

    LLM & Agent Capabilities

    • Multiple LLM Providers: Support for local providers (Ollama, LMStudio, Nativ, GPT4All) and cloud providers (OpenAI, Anthropic, Google, Grok, Mistral, Groq, Kimi, GLM, etc.).
    • MCP Support: Model Context Protocol servers for agent-like capabilities via a built-in Marketplace.
    • Agent Mode (v0.9.4+): Autonomous codebase exploration using read-only tools and parallel sub-agents.
    • Spec Driven Development (SDD) (v0.9.7+): Define tasks in Backlog.md and manage them via the Spec Browser (Task List and Kanban views) for autonomous implementation by the Agent.
    • ACP Runners (v0.9.10+): Communication with external agents via the Agent Communication Protocol (JSON-RPC 2.0).
    • CLI Runners (v0.9.9+): Execute prompts and spec tasks via external tools like Claude Code, GitHub Copilot, Codex, Gemini CLI, or Kimi.

    Coding Assistance

    • Inline Code Completion (v0.9.6+): AI-powered suggestions using Fill-in-the-Middle (FIM) models via Ollama or LM Studio.
    • Commands: Slash commands (e.g., /test, /explain, /review, /find) for common tasks.
    • Skills: LLM-activated capabilities loaded from SKILL.md files (compatible with .claude/skills and .agents/skills directories). Requires Agent Mode.
    • Git Diff/Merge: Integration to accept LLM suggestions via standard Git dialogs.

    Context & Interface

    • Project Scanner: Adds source code (full project or specific packages) to the prompt context.
    • Chat Interface: Features include local chat history storage, real-time streaming responses, and Drag-and-Drop (DnD) image support for multimodal models.
    • Web Search: Search capabilities using Google or Tavily.
    • Token Cost Calculator: Estimates costs when using Cloud LLM providers.
    • Appearance Customization: Adjust chat interface colors, spacing, and font sizes.

  10. What is DEVOXXGENIE.md and how does it work?

    master

    Starting from v0.5.0, DevoxxGenie supports a DEVOXXGENIE.md file located in your project root. This file acts as a structured context provider that is automatically included in the system prompt sent to the LLM at the start of a conversation.

    By providing this file, you enable the LLM to:

    • Understand your project structure and organization.
    • Adhere to your specific coding standards and practices.
    • Use the correct build system and commands.
    • Reference the appropriate dependencies and frameworks.

    This context allows the LLM to provide more tailored, accurate, and contextually relevant assistance compared to a generic chat session.

  12. What is RAG (Retrieval-Augmented Generation) in DevoxxGenie?

    master
    RAG allows the LLM to have accurate, project-specific context without manual file selection. DevoxxGenie indexes your source code into a local ChromaDB vector database (running in Docker) using Ollama embeddings. When you ask a question, the system retrieves the most semantically relevant code snippets and includes them in the prompt automatically.