jetbrains-ai capabilities
10 mapped capabilities, each graded and dated. This is the diagnosis — the migration guide is the cure.
Capabilities
Commit message generation and VCS assistance
provisionalverified 2 days agoAI Assistant generates commit messages from staged diffs, summarises changes, improves existing messages, and creates pull/merge request titles and descriptions. It can also flag potential issues via an AI Self-Review step before committing.
Data privacy controls and enterprise governance
provisionalverified 2 days agoJetBrains AI provides granular opt-in/out controls over data collection. Commercial and organisational users are opted out of detailed code-level data sharing by default; admins govern collection at the org level. Users can exclude files from AI context via .aiignore files.
In-editor AI chat (Chat / Agent / Codebase modes)
provisionalverified 2 days agoA persistent AI chat panel inside JetBrains IDEs supporting three modes: standard Chat for Q&A and code generation; Agent mode for multi-step autonomous tasks across files; and Codebase mode, which automatically gathers relevant project context via RAG-based retrieval before answering.
Inline AI code completion
provisionalverified 2 days agoReal-time ghost-text code suggestions powered by Mellum, JetBrains' proprietary code-optimised LLM. Suggests single lines, multi-line blocks, and entire functions as you type, across 28+ languages.
Junie CLI (standalone terminal agent, beta)
provisionalverified 2 days agoA standalone command-line evolution of Junie that runs outside any IDE — usable from a terminal, in CI/CD pipelines, or on GitHub and GitLab. It is LLM-agnostic, supporting models from OpenAI, Anthropic, Google, and xAI Grok.
Junie autonomous coding agent (in-IDE)
provisionalverified 2 days agoJunie is JetBrains' autonomous coding agent plugin for JetBrains IDEs. It operates in Code mode (plan, execute, test, self-correct across multiple files) or Ask mode (conversational codebase Q&A with full project context).
Local and offline model support (Ollama, LM Studio, BYOK)
provisionalverified 2 days agoAI Assistant can connect to locally-hosted models via Ollama or LM Studio, or to any OpenAI-compatible server (llama.cpp, LiteLLM, etc.), enabling AI assistance with no data leaving the machine. BYOK lets users supply their own API keys for Anthropic, OpenAI, Google, Vertex AI, or compatible endpoints.
MCP tool support (client and server)
provisionalverified 2 days agoJetBrains AI Assistant acts as an MCP client, connecting to external MCP servers to extend its capabilities with custom tools and data sources. Separately, JetBrains IDEs expose a built-in MCP server so that external agents (Claude Desktop, Cursor, etc.) can drive the IDE's code analysis, debugging, and refactoring tools.
Next Edit Suggestions
provisionalverified 2 days agoPredicts the next code location you are likely to edit and proposes the change inline. Differs from completion by modifying existing code rather than inserting new text, and can propose edits beyond the immediate cursor position.
Plans, credits, and usage gating
provisionalverified 2 days agoJetBrains AI is sold in four tiers — AI Free, AI Pro, AI Ultimate, and AI Enterprise — using a credit-based quota system where 1 AI Credit equals $1 USD. Different features consume credits at different rates; inline code completion and Next Edit Suggestions availability vary by tier.