Cursor brings AI agents to JetBrains IDEs via ACP

Cursor has just expanded into JetBrains IDEs through the Agent Client Protocol. Developers can keep IntelliJ workflows while adding Cursor agents, model choice across major providers, plus secure indexing and semantic search inside the IDE.

cursor cover

TL;DR

  • JetBrains IDE support via ACP: Cursor available in IntelliJ IDEA, PyCharm, WebStorm, and other JetBrains IDEs
  • Keeps JetBrains workflows: Adds Cursor’s agent-driven capabilities alongside existing JetBrains code intelligence and tooling
  • Model flexibility: Choose models from OpenAI, Anthropic, Google, and Cursor via https://cursor.com/docs/models
  • Per-model agent harness: Custom-built harness per model to tune quality and performance across coding tasks
  • Secure indexing + semantic search: Brings Cursor’s indexing and semantic search into JetBrains for large codebase understanding
  • Setup + pricing: Install Cursor ACP in JetBrains AI chat; authenticate; free for paid plans

Cursor is expanding beyond its own editor experience: the company says Cursor is now available inside IntelliJ IDEA, PyCharm, WebStorm, and other JetBrains IDEs by way of the Agent Client Protocol (ACP). The net effect is straightforward: teams that live in JetBrains for Java and multilanguage work can keep their existing IDE workflows while adding Cursor’s agent-driven approach on top.

Cursor agents, now living in JetBrains

Cursor’s JetBrains integration centers on running Cursor through ACP, which is positioned as a way to bring ACP-compliant agents into JetBrains IDEs. Practically, that means Cursor’s agent functionality is surfaced in the same environment where JetBrains developers already rely on deep code intelligence and tooling.

Cursor frames the benefit as “agents effective across all surfaces,” but the key developer-facing detail is that ACP lets Cursor sit alongside JetBrains’ IDE capabilities rather than replacing them.

Model choice via ACP: OpenAI, Anthropic, Google, and Cursor

A notable part of the integration is model flexibility. With Cursor ACP, developers can explore and choose from “frontier models” listed in Cursor’s documentation, including options from OpenAI, Anthropic, Google, and Cursor itself (via Cursor’s model docs).

Cursor also says its agent harness is custom-built for every model to optimize output quality and performance—an implementation detail that matters in day-to-day use because different models tend to behave differently across tasks like code navigation, refactors, and multi-step changes.

Codebase indexing and semantic search, inside the IDE

Cursor is also bringing its secure codebase indexing and semantic search into the JetBrains environment. The idea is to help the agent understand large codebases and then pair that understanding with JetBrains’ existing code intelligence.

For enterprise-scale repositories in particular, the combination of indexing, semantic search, and IDE-native tooling is the core technical promise of this integration.

Getting started

Setup runs through JetBrains’ AI chat: install the Cursor ACP, then authenticate using an existing Cursor account. Cursor says the Cursor ACP is free for all users on paid plans, with more details in its documentation at cursor.com/docs/cli/acp#ide-integrations.

What’s next

Cursor describes ACP as a foundation for deeper JetBrains integrations, with the goal of bringing more agentic coding capabilities to JetBrains users over time.

Source: https://cursor.com/blog/jetbrains-acp

Continue the conversation on Slack

Did this article spark your interest? Join our community of experts and enthusiasts to dive deeper, ask questions, and share your ideas.

Join our community