Junie

Junie is an autonomous terminal-based AI coding agent developed by JetBrains. Built on top of the IntelliJ IDEA engine, it combines semantic code intelligence and runtime capabilities with multi-model AI reasoning across local terminal environments, GitHub Actions, and GitLab CI/CD pipelines. In day-to-day development, Junie runs as a CLI orchestrator that breaks complex objectives into structured stages. It allows developers to configure separate models for different phases—such as using high-reasoning models for planning and faster models for code generation. Users can interact via live prompting to redirect the agent mid-task without resetting context, and can connect Junie to over 10 model providers (including Anthropic Claude, OpenAI, Google Gemini, and Grok) or local models via Bring Your Own Key (BYOK). What sets Junie apart from standard CLI agents is its agentic debugging and IDE-grade runtime integration. Instead of merely analyzing text logs, Junie can start actual debug sessions, place breakpoints directly in source code or compiled bytecode/JARs, evaluate runtime expressions, and step through execution traces. It also stores editable, version-controlled execution plans in `.junie/plans` and supports remote monitoring across devices. Junie is available with free initial credits and allows ongoing zero-markup BYOK usage. JetBrains also offers subscription tiers (AI Pro and AI Ultimate) with bundled cloud credits and SOC 2 security compliance.

Tags: ai agent, cli, code assistant, coding agent, terminal

Visit Junie

Key features

  • IntelliJ Engine & Agentic Debugging — Attaches to live debuggers, sets breakpoints in source and compiled JARs, and inspects runtime stack frames. - Dual-Model Plan & Execution — Generates committable markdown plans in .junie/plans with high-reasoning models before executing on faster models. - Live Prompting & Dynamic Steering — Interactively adjusts agent decisions mid-task without restarting the execution context. - BYOK & Multi-Model Support — Direct integration with 10+ cloud model providers (Claude, GPT, Gemini, Grok) and local LLMs at provider rates. - CI/CD Integrations — Executes directly within GitHub Actions and GitLab CI/CD pipelines to triage issues and author pull requests. - Model Context Protocol (MCP) — Connects to external data sources, databases, and APIs using the standard Model Context Protocol. - Remote Control Monitoring — Triggers long-running tasks via CLI and monitors progress asynchronously via a web interface.

Use cases

Use Case: Deep Runtime Debugging Diagnose complex failures where static analysis fails by letting Junie launch a real debugger, navigate stack frames, step through external libraries or bytecode, evaluate runtime variables, and implement code fixes.

Use Case: Multi-Step Framework Migrations Execute large-scale dependency updates (such as framework major versions) by running an initial plan stage, approving architectural changes in .junie/plans, and allowing the agent to refactor code and verify test suites asynchronously.

Use Case: Automated CI/CD Issue Resolution Configure Junie within GitLab CI/CD or GitHub Actions to automatically pick up reported bug tickets, replicate the problem with new test cases, implement fixes, and submit draft merge requests for maintainer review.

Who it is for

Junie is designed for professional software engineers, backend developers, and teams using JVM or multi-language tech stacks who require deep runtime debugging and structured execution planning from the CLI. It is less suitable for users looking for lightweight single-line autocomplete or developers who do not need full-agent project orchestration.

Install guide

Installation Install Junie on macOS, Linux, or Windows using the official installation script: bash curl -fsSL https://junie.jetbrains.com/install.sh | bash

First Run Initialize Junie in your project workspace and configure your preferred model provider keys: bash junie init junie config set-key anthropic <YOUR_ANTHROPIC_API_KEY>

Example Session Generate an execution plan for a refactoring task and run it with human-in-the-loop review: bash junie plan "Refactor authentication middleware to use JWT validation" cat .junie/plans/auth-refactor.md junie run --plan .junie/plans/auth-refactor.md

Part of CLI Coding Agents

Categories: AI Coding Agents, Code Assistants, Developer Tools

Related CLI agents

  • Claude Code — Claude Code is an official terminal-based agentic coding tool developed by Anthropic. Executed via the `claude` CLI command, it provides…
  • Keen Code — Keen Code is an open-source, terminal-based AI coding agent built in Go by developer mochow13. It runs directly inside shell…
  • FetchCoder — FetchCoder is a terminal-based AI coding assistant developed under the Fetch.ai organization. Built specifically for command-line workflows, the tool provides…
  • Kolega Code — Kolega Code (kolega-code) is an open-source Python-based terminal coding agent developed by the Kolega AI team. It is built to…
  • ZAP — ZAP is an open-source, terminal-based AI coding agent implemented in Rust by the zap-coding-agent project. It provides an autonomous command-line…
  • Open Codex — Open Codex is an open-source, community-maintained fork of OpenAI's Codex CLI tooling, maintained by ymichael on GitHub. It provides a…
  • AI Tools
  • Categories
  • Industries
  • CLI Coding Agents
  • MCP Servers
  • MCP Categories