On June 17, 2026, JetBrains moved Junie out of beta. Junie is the company's AI coding agent for IntelliJ IDEA, WebStorm, PyCharm, and the rest of the JetBrains IDE family, plus a standalone Junie CLI for terminal, CI/CD, and GitHub/GitLab workflows. The GA release is not a rebrand: JetBrains rebuilt the IDE integration on ACP (Agent Communication Protocol), added agentic debugging through the real IDE debugger, and hardened the features teams said they needed before trusting an agent on production codebases.
What Junie Actually Does
Junie is an LLM-agnostic coding agent. You give it a high-level task in natural language; it plans, edits across the project, runs terminal commands, uses MCP servers, and reports progress. Unlike a chat sidebar that suggests snippets, Junie is built for multi-step work: migrations, feature implementation, test fixes, and PR review with project context loaded.
Authentication options at GA: JetBrains Account (OAuth), a Junie API key, or Bring Your Own Key from Anthropic, OpenAI, Google, xAI, OpenRouter, or GitHub Copilot. You can also point Junie at local runtimes (Ollama, LM Studio, LiteLLM) so prompts and code never leave your machine.
Agentic Debugging: The GA Headline
The feature JetBrains leads with is debugging that uses the same debugger you do. Instead of sprinkling println and guessing, Junie can set breakpoints, step through execution, inspect variables, and iterate on fixes against live runtime state. That matters for web developers chasing race conditions, stale React state, or API contract mismatches where static code review alone fails.
This is the differentiation JetBrains has been telegraphing since the experiment started: an agent that does not guess at project details but uses the IDE's actual tooling (debugger, test runner, refactorings, semantic search) the way a senior engineer would.
ACP and Deep IDE Integration
Junie's GA IDE bridge runs on ACP, the same protocol Junie CLI uses to talk to your editor. Practical effect: CLI sessions and IDE sessions share context semantics. Start a migration on your laptop, monitor progress from your phone, review the resulting PR later. Junie CLI also connects back into a running JetBrains IDE for shared context and semantic references.
Junie supports MCP server connections for filesystem access, issue trackers, and custom tools. Combined with JetBrains' semantic code model inside the IDE, the agent gets structural understanding (classes, imports, usages) rather than treating the repo as a flat text dump.
Code Review and Long-Running Tasks
GA Junie can review pull requests with project conventions in scope: naming rules, review checklists, and team standards you teach it persist across sessions. Long-running tasks run asynchronously while you work on something else, with remote monitoring so you are not babysitting a terminal tab.
Where Junie Sits in the Stack
JetBrains now has three layers for agentic development:
- Junie — the agent that plans, codes, debugs, and reviews inside your IDE or CLI.
- JetBrains Central — the team control plane for governing agents from any vendor.
- JetBrains AI subscription — bundled cloud credits, chat, and Junie access tiers (Free to Start, AI Pro, AI Ultimate).
Against Cursor 3 and GitKraken Kepler, Junie bets on IDE-native depth over a greenfield editor or a merge-focused ADE. If your team already lives in WebStorm or IntelliJ, GA Junie is the path of least resistance. If you mix agents across repos and providers, Central plus Junie CLI is the JetBrains answer to vendor lock-in, consistent with what their 90% adoption survey found developers demand.
Pricing and Availability
Junie is available in all JetBrains IDEs and through Junie CLI on macOS, Windows, and Linux. Existing JetBrains AI subscribers get access through their plan; BYOK users pay provider rates with zero markup from JetBrains. Junie Free to Start covers exploration; AI Ultimate is positioned for regular agent workloads with higher cloud credit limits.
If you have been waiting for an IDE-native agent that debugs like you do rather than vibes like a chatbot, GA is the release to trial on a real bug first, not a toy kata.