Junie

What is Junie?

Junie is JetBrains' own AI coding agent, built into the company's IDE family (IntelliJ IDEA Ultimate, PyCharm, WebStorm, GoLand, PhpStorm, Rider, RubyMine, RustRover) and also shipped as a standalone Junie CLI for the terminal, GitHub Actions and GitLab CI. It reached general availability on June 17, 2026, after roughly a year of internal development.

Rather than lock users to one model, Junie reaches frontier models from Google, OpenAI, Anthropic and xAI through a JetBrains AI subscription's shared credit pool, a bring-your-own API key for OpenAI or Anthropic, or a fully local runtime via Ollama, LM Studio or on-premises LLMs. As of August 2026 the everyday default is Gemini 3.7 Flash, at 40% off base pricing for a limited time.

JetBrains itself is a 25-year-old, privately-held company headquartered in Prague, grown without venture funding, best known for IntelliJ IDEA, PyCharm and Kotlin. Junie's built-in base of professional IDE users gives it a head start most venture-funded coding-agent startups do not start with.

What does Junie do?

From inside a JetBrains IDE or the standalone Junie CLI, you describe a task and Junie first produces a Plan, a structured document covering requirements, technical design and delivery stages, that you review and approve before any code gets written. Once approved, it edits files, runs and inspects tests, and for debugging tasks it drives the IDE's own debugger directly.

It sets breakpoints and inspects live runtime state rather than inferring behavior from log output. Long-running tasks execute asynchronously in the background while you keep working elsewhere. Junie can review a pull request with full project context, including build config, tests and team conventions, and through a DataGrip connection it can work directly against a configured database for SQL-related tasks.

By default the agent runs on Gemini 3.7 Flash at 40% off base pricing, with OpenAI, Anthropic, Google and xAI also reachable through JetBrains' own credit pool or a bring-your-own API key.

Outside the IDE, the same agent runs via the Junie CLI in a terminal, in GitHub Actions (triggered from an issue or PR comment via JetBrains' official Junie GitHub Action), or in GitLab CI, so it can be scripted into a pipeline instead of only run interactively.

Organizations that join the JetBrains Central governance layer, rolling out through summer 2026, can run Junie under the same shared context and 12-month AI Credits pool that JetBrains Central uses to manage Claude Code, Codex and Gemini CLI workflows.

How Junie works

1. 1Install the Junie plugin (bundled into recent JetBrains IDE versions) or the standalone Junie CLI for terminal, GitHub Actions or GitLab CI use.
2. 2Describe a task. Junie's Plan Mode produces a structured requirements, technical-design, delivery-stages and testing-strategy document for you to review and approve before any code is written.
3. 3Once approved, Junie edits files, runs commands and tests, and for debugging tasks drives the IDE's real debugger (setting breakpoints and inspecting live runtime state). By default the agent runs on Gemini 3.7 Flash at 40% off base pricing.
4. 4Long-running tasks execute asynchronously in the background while you keep working. Junie can also review a pull request with full project context or work against a DataGrip-connected database.
5. 5Every run draws from your JetBrains AI subscription's shared credit pool, a bring-your-own OpenAI or Anthropic API key, or a fully local model via Ollama, LM Studio or on-premises LLMs. Organizations can join the JetBrains Central layer for shared governance and a 12-month AI Credits pool.

Who is Junie for?

• ✓Developers on IntelliJ IDEA Ultimate, PyCharm (with a Pro subscription), WebStorm, GoLand, PhpStorm, Rider, RubyMine or RustRover who want an agent that shares the same debugger and project index rather than a separate tool bolted on top
• ✓Teams that want to review a written plan before code changes happen, not just receive an autonomous diff
• ✓Organizations using the JetBrains Central governance layer (rolling out summer 2026) to monitor and budget shared AI Credits across developers and tools
• ✓Anyone who wants to point the same agent at frontier models from Google, OpenAI, Anthropic or xAI through JetBrains' own credit pool, a bring-your-own API key, or a fully local model via Ollama, LM Studio or on-premises LLMs

Not forDevelopers and teams not using a JetBrains IDE day to day. Junie's CLI, GitHub Action and GitLab CI integration work standalone, but its two signature features (Plan Mode's UI and Agentic Debugging) are built into the JetBrains IDE experience and don't carry over to VS Code, Vim or another editor. Also teams that want a single flat per-seat price rather than a shared, feature-wide credit pool.

Why use Junie?

1. Agentic Debugging drives the IDE's real debugger. It sets breakpoints and inspects live runtime state, instead of guessing from logs the way a terminal-only agent has to.
2. Plan Mode produces a reviewable requirements and design document before any code is written, rather than jumping straight to a diff.
3. Gemini 3.7 Flash became the default model in August 2026 at 40% off base pricing, with Google, OpenAI, Anthropic and xAI frontier models also reachable via JetBrains' own credit pool, BYOK, or a fully local Ollama or LM Studio runtime.
4. Backed by a 25-year-old, privately-held, profitable software company with an already-large professional IDE install base, not a VC-funded startup burning runway.
5. Organizations that join JetBrains Central (rolling out summer 2026) get shared context, governance and 12-month AI Credits pooling across Junie and other agents their developers use, including Claude Code, Codex and Gemini CLI.