PyCharm 版本發佈

Key highlights

• Python extension development with the Rust plugin (Beta):
  • Work seamlessly with Python projects that leverage Rust modules to speed up performance-critical components.
  • Jump between Python and Rust with bidirectional navigation from .pyi stubs to their underlying pyo3 code and vice versa.
  • Automatically rebuild Rust extensions during development whenever you run or test the Python application.
  • Debug compiled Rust extensions natively alongside your standard Python debugging workflows.
• 'debugpy' as the default debugger - Following its introduction as an optional backend in 2026.1, debugpy is now enabled by default for all Python projects and Jupyter notebooks, using the Debug Adapter Protocol (DAP).
• Support for uv-backed tools and uvx - PyCharm now leverages the uv toolchain to streamline how you run your external development utilities, eliminating manual package setups that clutter your local environment. The settings interface for external tools and linters - such as ruff and black - has been completely redesigned. Additionally, you can now configure these utilities to run via uvx when they are not installed in your local project environment. This allows you to manage tool integrations at the project level or establish defaults for all new projects, while executing the underlying packages globally on your machine. You can now run uv commands directly within the built-in terminal inside your project environment.
• Support for uv, Poetry, and Hatch multi-projects and uv workspaces (Beta) - Previously available as an optional feature in PyCharm 2026.1.1, this functionality is enabled by default in version 2026.2. It streamlines your subproject management and provides richer dependency insights directly within your configuration files. See the exact installed version of a package as a gray inline hint displayed directly next to its requirement definition in pyproject.toml or requirements.txt. Hover over a package name or press Ctrl+Q to view information about the package, including current version, license, and upstream repository links. Hold Ctrl and click on a package name to open its specification in an editor tab, or navigate directly to the source module in the Project tool window for internal workspace dependencies.
• Editor minimap - Navigate complex source files and notebooks more efficiently with the official editor minimap. It provides a high-level visual overview of your document structure across all supported file types - while offering a dedicated layout built just for Jupyter notebooks. Quickly jump to specific cells or sections in long notebooks by clicking directly on the minimap, reducing scrolling time. Hovering over the minimap automatically highlights the corresponding code block, making it easier to track your position in large files. Position the minimap on either side of the editor, or choose between the Fit and Fill scaling modes to optimize how the document view is displayed. Toggle the minimap by searching for it via Search Everywhere (Shift+Shift) or use the three-dot menu in the upper-right corner of the editor.
• Pyrefly type engine integration - Use Pyrefly as an external type engine to significantly accelerate code insight features for large-scale Python codebases. Leverage Pyrefly's Rust-based architecture for faster type inference and diagnostic resolution compared to the built-in engine. Improve the precision of type-related diagnostics, inlay hints, and quick documentation by delegating analysis to the Pyrefly language server. Switch to the Pyrefly engine via the Type widget in the status bar at the bottom of the IDE. PyCharm will automatically handle the installation if it's not already configured. Note: This integration is currently optimized for local interpreter configurations. Support for Docker, WSL, SSH, Jupyter Notebooks, and multi-module projects is planned for upcoming releases.

AI in PyCharm

• Start new projects with AI - If you have a JetBrains AI license, you can now generate fully configured, runnable projects from scratch using natural language prompts directly from the Welcome screen. Generate entire codebases from natural language, including boilerplate code, configuration files, and database connections based on a plain-text description. Skip manual environment setup as the IDE automatically configures the correct SDK, installs required dependencies, and creates functional run configurations. Refine code iteratively through a persistent chat thread that lets you add features or modify the generated artifacts on the fly.
• Agent skills manager - AI agents are only as useful as the context they have. When they don't have knowledge of your frameworks, conventions, and tooling, you end up re-explaining the same setup in every new chat window. Agent skills fix that. Install them once in PyCharm, and your agents carry that domain knowledge across every project and session - automatically. Browse and manage skills directly from the IDE, expand the built-in library with external registries like public GitHub repositories, or let PyCharm import skills you've already set up for Claude Code or Codex.
• Deprecation of the Machine Learning Code Completion (Ranking) plugin - JetBrains are deprecating the Machine Learning Code Completion (Ranking) plugin. As AI completion, next edit suggestions, and agents have become the primary ways developers get intelligent assistance, demand for ML-based completion ranking has significantly declined. The plugin will no longer be bundled, but it remains available for installation from JetBrains Marketplace if you'd like to continue using it.

Code insight

• Improved literal type narrowing ensures that the IDE correctly infers Literal types when assigning values to variables, preventing unnecessary type warnings.
• Enhanced Pydantic code insight and inspections deliver a more reliable experience when working with Pydantic models. Available in PyCharm Pro.
• Support for PEP 561 partial stubs ensures you get seamless code completion and accurate import suggestions. You can now effortlessly discover and import modules that aren't explicitly declared in a library's stubs, as PyCharm automatically falls back to the underlying source implementation or typeshed to find them.

New Welcome screen available for more users

• The non-modal Welcome screen is now being rolled out to a wider audience, providing a smoother start to your workflow by reducing visual clutter. It now appears as a dedicated tab inside the IDE, keeping your most common actions within easy reach.

Platform

• Streamlined Git conflict resolution flow - In large projects, merges often result in dozens of file conflicts, many of which are simple import mismatches or formatting differences. PyCharm 2026.2 improves the conflict resolution flow, allowing you to invoke the Resolve All Simple Conflicts action, which automatically resolves standard conflicts across the entire changeset instantly, instead of having to open each file one by one. With the new flow, you can also see a complete overview of resolved and unresolved files, reopen files at any time, and revert and reapply conflict chunks in any order.
• Easier file and image sharing in terminal sessions - No more copying and pasting paths from your project or external applications when working in the terminal or with CLI agents. You can now drag and drop a file or folder directly into the terminal to insert its path into the current command. To quickly start working from a specific location, drop a file or folder onto the terminal tab area to open a new tab with its root path already set. You can now also paste images directly from the clipboard into supported CLI agent sessions, which makes it easier to share visual information with your agent.

Web

• TypeScript 7 support for faster workflows with large codebases - Large TypeScript codebases don't have to feel slow. TypeScript 7 rewrites the compiler and language server in Go, and the difference is noticeable immediately. In testing, project load time on the Kibana codebase dropped from ~12 seconds to ~3 seconds, which means editor responsiveness has improved roughly fourfold. PyCharm 2026.2 supports TypeScript 7 as the default for projects already using it, with an opt-in upgrade path for those still on earlier versions. You get the full speed benefit immediately - no full project migration required on day one. For teams on large TypeScript monorepos, this is a day-one upgrade.

Cloud

• Docker Compose statuses right in the editor - Docker Compose files are now more informative and interactive. You can view the status of running containers directly inside your Docker Compose file and quickly access common actions without leaving the editor. With the ability to jump straight to logs for a specific service, create a database connection, or open a service in your browser directly from the corresponding entry in the Compose file, it is easier to monitor and work with your application stack.
• Terraform testing framework support - Version 2026.2 extends PyCharm's infrastructure-as-code capabilities by introducing native support for the Terraform test framework. This integration rounds out the IDE's ecosystem support, allowing you to validate module behavior and enforce configuration policies early in the development lifecycle. These additions build on out-of-the-box infrastructure tooling, ensuring your deployment configurations remain highly maintainable, predictable, and clean.

• Code Insight
  • Pyrefly: false positive: 'Getter signature should be (self)'.
  • Concatenate not taken into account for variance inference.
  • Empty list literal collapses generic unification to 'Any'.
  • Passing 'None' to generic doesn't work.
  • Can't change type engine when looking at .pyi file.
• Jupyter
  • Markdown cell table alignment: Columns not aligning in Jupyter Notebook.
• LSP Tools
  • Python LSP plugin regression: Pyright and Ruff diagnostics not working after latest update on 2026.1.2.
• Restructured Text
  • Python code block in docstring does not render.

• You can now infer function parameter from decorator.Improved syntax highlighting for f-string format specifiers.
• Added support for bounds and variance for type variable tuples and parameter specifications.
• Added inlay hint for number promotions.
• LSP Tools
• Now suggests turning ruff on if detected in the environment.

• BDD
  • "Undefined step" Cucumber inspection flags valid pytest-bdd multi-decorator steps as ambiguous.
  • Behave runner deletes C-extension modules from sys.modules between dry-run and run, crashing scipy on reimport.
  • Behave support: ImportError: No module named machinery.
• Code Insight
  • PyCharm shows class-decorated functions in the import menu as functions, not as values.
  • Pyrefly: false positive Getter signature should be (self).
  • Updated typeshed and django stubs.
  • False errors in Annotated metadata parameters
  • "Non-idempotent computation" - PySuspiciousBooleanConditionInspection.
  • Passing None to generic doesn't work.
  • pyrefly: wrong module import suggestions.
  • Flaky false positive Member 'None' of 'X | None' does not have attribute ... for variables of X type.
  • Can't change type engine when looking at .pyi file.
• Databricks
  • Databricks plugin: BootstrapMethodError when testing connection after 2026.1 update.
• Debugger
  • Python type renderers don't seem to work at all (pydevd).
  • [debugpy] No option for 'View as Dataframe' for evaluated dataframes.
  • [debugpy]: Custom CA signed TLS certificate verification fails: [SSL: CERTIFICATE_VERIFY_FAILED].
• Usability
  • [debugpy] Popups about debugpy availability shouldn't show when remote interpreter is selected.
• Editing
  • Double click f-string quote character to select entire string.
• Inspections
  • Incorrect suggestion in PyCharm expression simplification hint.
  • False positive: Type 'UUID' doesn't define 'str' or 'repr', so the result won't be useful.
  • False positive: Type 'Decimal' doesn't define '__str__' or '__repr__'.
• Jupyter
  • jetTransient property in notebook causes nbdiff to fail validation.
  • Jupyter not showing output in dev container.
  • Ipywidgets aren't rendered in Jupyter notebooks via Remote Development.
  • The Bokeh-generated graph cannot be displayed in Jupyter within PyCharm 2025.3.3.
  • Jupyter: Fix table coroutine scope.
  • Jupyter notebook layout breaks in PyCharm after unaccepted Cursor changes.
  • Jupyter: Remote dev
  • RemDev: Interactive matplotlib plots don't work in Jupyter Notebooks.
• Remote Interpreters
  • uv run is not working for remotes.
  • Tables (Jupyter, DataView)
  • Incorrect Missing Values Count in Pandas DataFrame Preview.

• AI Toolkit
  • [AI Playground] Detect and use system prompt when pasting user prompt.
• Code Insight
  • Introduced 'PyAnyType'.
  • Added support for a tuple with test cases for 'pytest.mark.parametrize' inlay hints.
  • Python: Improve UI for type engine.
  • Updated django stubs.
  • Added inlay hints for pytest, function parameters and number promotions.
• Debugger
  • debugpy: added support for multiple path mappings.
• IDE
  • Bundle Code provenance plugin.
  • Added support for multiple virtual environments (via pipenv Pipfile) per project.
• Inspections
  • Added inspection for unhelpful string operation on type.
• Interpreters Settings
  • Activate interpreter/venv on a per file/folder basis in PyCharm.
  • Added hatch run configuration.
  • Allow separate interpreter selection for different paths.

• BDD
  • "Undefined step" Cucumber inspection flags valid pytest-bdd multi-decorator steps as ambiguous.
  • Behave runner deletes C-extension modules from sys.modules between dry-run and run, crashing scipy on reimport.
• Code Insight
  • PyCharm shows class-decorated functions in the import menu as functions, not as values.
  • PyCharm shows overloads as unions, which is incorrect.
  • Type Checker: Literals do not work with dataclass Keyword arguments.
  • Use weak keys for TypeEvalContext caches.
  • Narrowing X | None to X with if x is None: x = X no longer works with nightly PyCharm PY-261.18125.
  • False positive: Cannot use unhashable type 'list' as a set element.
  • 'async def' should return a types.CoroutineType, not a typing.Coroutine.
  • Skip parameter inlay hints for self and cls.
  • State type engine is beta in the UI.
  • Adjust CSP solution to use correct default type.
  • 'fixme' test utility improvements.
  • Type var solved to default when it should remain unsolved.
  • Slight improvement of debuggability in PyTypingTypeProvider.
  • Partially unresolved union leads to complete 'Any'.
  • PyCharm complains about about using type[bytearray] as set element - unhashable type.
  • False positive Expected type 'X', got 'EllipsisType' instead passing ellipsis as default value in overloads.
  • False negative using None default value for a non-optional parameter.
  • Disallow using external type checkers with remote interpreters.
  • Iterating over a collection of collections.namedtuples considers elements to be tuple[Any, ...].
  • 2026.1 regression: no error when incorrect type in generic function.
  • pyrefly: enabling external type engine sometimes fails with ISE: Cannot find LSP executable file.
  • No warnings with built-in type engine after creating a project with external one.
  • False errors in Annotated metadata parameters.
• Databricks
  • Databricks plugin: BootstrapMethodError when testing connection after 2026.1 update.
• Debugger
  • Function is executed on hover while debugging.
  • debugpy: cannot debug tox - Server is not available.
  • [debugpy]: debugger doesn't stop on exceptions if "On Raise" is enabled.
  • debugpy populates PYTHONPATH in the run configuration.
  • Evaluate expression does not work PyCharm 2026.1 EAP 5 with debugpy (Mac, arm).
  • Chinese character garbled in Debugger Console with debugpy.
  • [debugpy] Debugger doesn't stop on breakpoints in django templates (HTML and XML).
  • debugpy: debugger stops in _jb_pytest_runner if pytest fails.
  • debugpy: full values are not shown in debugger.
  • [debugpy] Debugger does not inject the configured environment variables from env file.
  • Attach to DAP: debug button is disabled if the pydevd backend is chosen.
  • The debugger stops on code update (Process finished with exit code 3).
  • [debugpy] Debugger backend switcher is not visible in Current File configuration.
• Django
  • Django server won't auto restart.
• Docker
  • Unable to add docker compose interpreter within Python 2.7 in 2026.1 for Linux.
• Inspections
  • Type error sorting Iterable[] of dataclass(order=True) instances.
  • False positive: Type 'Path' doesn't define 'str' or 'repr', so the result won't be useful.
• Interpreters Settings
  • IDE Zoom causing misalignment on the New Project Wizard.
  • Can't create new or select existing virtual environment based on Pyenv of Mise.
  • In the quick fix for configuring a module interpreter replace "project" with more specific wording.
  • Adding existing poetry env with pyproject not in project root.
  • Adding a python sdk to a module always uses the project root instead.
  • Make 'hatch run' the default run configuration for Hatch SDKs.
  • Thread pool exhaustion in PyCharm backend when trying to access a password for a private PyPI repository.
  • POTW - Rewrite process output to be more console-like.
  • POTW - Process tree autoscroll breaks due to a cancelled collector.
  • POTW - Some processes remain unfinished indefinitely.
  • Interpreter setup blocks me from working.
• Jupyter
  • Variables Disappear from Jupyter Notebook Variable Explorer When Loading CLIP Objects.
  • Jupyter debugger does not highlight the current line.
  • New Colors for Matplotlib Diagrams in Jupyter Notebooks.
  • Jupyter Notebook Debugger does not work for a scratch file.
  • "Reformat Code" popup lingers for 30 seconds every time Jupyter notebook is saved.
• Jupyter: Debugger
  • [WSL] Step Into doesn't work in Jupyter Debugger.
• Packaging
  • Support poetry based environments when the pyproject.toml is not in the root folder (e.g. multiple projects in single repo).
  • UV does not correctly sync or add packages.
  • UV python manager version control wrong version highlight.
  • Python packages toolwindow doesn't refresh after performing the 'pipenv update' action.
  • Hide Python Process Output and Python Packages from UI panel of other IDEs, if there are no Python modules in the project.
  • Extras in pyproject.toml dependencies parsed as version constraints.
• Quick Documentation
  • PyCharm leaves unknown Sphinx roles in the rendered docstrings.
• Remote Interpreters
  • SSH connection issues triggered when editing python interpreter.
• Run
  • Excluding folders lost after reopening projects with source root in 2026.1.
  • File/Folder URI in output always opens browser instead of file manager/related application.
  • Cannot Edit or Create Run configurations, also edit Python tools.
  • Macros such as $FilePath$ stopped working in Run/Debug Configuration for 2026.1.
  • Rewrite pyproject.toml workspace sync to use direct mutation.
• No subsystem
  • No python configured for the newly created project on clean pycharm instance on Mac.
  • [FUS] Remove cache workaround for pyproject collector after race condition is fixed.
  • PTW instant update of sdk name after switching sdk.

Support for debugpy as an option for the default debugger

• debugpy is now available as a debugger backend option for local development, providing the industry-standard Debug Adapter Protocol (DAP) and a more stable connection model.
• The new engine provides full native support for PEP 669, utilizing Python 3.12's low-impact monitoring API to significantly reduce overhead.
• Delivers comprehensive asyncio support, allowing you to use the debug console and expression evaluation directly within asynchronous contexts.
• Provides a streamlined and reliable Attach to Process experience for local workflows, Docker containers, and remote servers. For specialized workflows, a new Attach to DAP run configuration allows you to connect to targets using debugpy.listen(), eliminating the friction of manual connection management.

Support for uv as a remote interpreter

• Introduces uv support for remote targets allowing developers to work on projects where code and dependencies live on a remote server - whether via SSH, in WSL, or inside Docker these environments stay synchronized.
• Use uv as your remote interpreter to manage packages through the Python Packages tool window and enjoy automatic Jupyter server launches just as if you were working locally. This integration removes the friction of manual environment orchestration across different infrastructures, letting you focus on your code instead of managing your remote setup.

Google Colab support

• Google Colab support is available as a core feature, along with basic Jupyter Notebook support. If you already use Google Colab, you can now bring your notebooks into PyCharm and work with them using IDE features designed for larger projects and longer development sessions.

JetBrains AI

• ACP Registry in PyCharm - In addition to Junie, Claude Agent, and Codex, PyCharm now lets you work with more AI agents directly in the AI chat. Choose from GitHub Copilot, Cursor, and many others supported through the Agent Client Protocol (ACP).
• Discover available agents and install them in just one click with the new ACP Registry.

Jupyter notebooks

• Opening notebooks directly in the IDE without a project - You can now open Jupyter notebooks (.ipynb) directly from your system file explorer in PyCharm. These files open within the Welcome screen project, which automatically indexes the notebook and applies inspections using your global Python interpreter. This allows you to run cells, install missing packages via quick-fixes, and use advanced coding assistance immediately without the need to create or configure a formal project. This functionality also supports a wide variety of the most used file extensions:
  • Python and scripts: .py, .sh.
  • Data and configuration: .json, .yaml, .yml, .toml, .env, .csv, .tsv, .xml, .sql.
  • Documentation and web: .md, .txt, .html.
  • DevOps: Dockerfile.
• A new debugger option - The Jupyter debugger now also has an option to use DAP as the default debugger, ensuring a consistent and more stable debugging experience across all your Python projects.

Productivity-enhancing features

• Ability to work with projects from Dev Containers as with local ones - Modern software development increasingly relies on Dev Containers for defining consistent, professional-grade environments. While PyCharm has long supported these via Remote Development mode, a more streamlined approach is now available to keep you in the flow. Starting with 2026.1, you can open projects from a Dev Container natively in your local IDE. Unlike previous methods that required a full IDE backend inside the Dev Container, this new agent-based architecture provides a high-performance connection to the file system without draining system resources. This reduces cognitive load and setup time, ensuring you stay focused on your most important tasks.
• Editor caret and selection updates - Smooth caret animation and updated selection behavior provide improved comfort, a cleaner look, and a more enjoyable coding experience.
• In-terminal completion - Stop memorizing commands. Start discovering them. In-terminal completion helps you instantly explore available subcommands and parameters as you type. Whether you're working with complex CLI tools like Git, Docker, or kubectl or using your own custom scripts, this feature intelligently suggests valid options in real time. Previously introduced for Bash and Zsh shells, it is now also available in PowerShell.

Web technologies

• JavaScript and TypeScript - Experience a smoother, more reliable workflow with TypeScript support now running on the service-powered type engine by default. Optimized for professional-grade development, this update delivers more accurate type inference and significantly lower CPU usage, even in large-scale enterprise projects. The TypeScript support is further improved with better auto-import handling for path aliases and project references, as well as the integration of inlay hints from the Go-based TypeScript language server. JavaScript parsing now also correctly handles string-literal import / export specifiers. PyCharm keeps you at the cutting edge of the web ecosystem with refined support for your favorite frameworks:
  • The IDE now highlights React's new use memo and use no memo directives.
  • The Vue integration uses the updated 3.1.8 version of @vue/typescript-plugin.
  • Astro settings accept JSON-based configuration for language server integration.
  • Angular 21.x template syntax is supported.
  • Django 6 support - PyCharm now supports Django 6, including full coding assistance for template partials via {% partialdef %} and {% partial %} tags. The update provides autocompletion, live template rendering, and seamless navigation between partials and their corresponding views or block implementations. Additionally, you can now run and debug Django 6 background tasks directly using gutter icons near the task decorator.
• FastAPI dev Run mode - You can now toggle this mode via a new checkbox next to the Run using options in the Run/Debug Configurations dialog. This update allows you to easily switch between server launch modes while maintaining access to existing uvicorn or Hypercorn parameters.
• Celery integration - PyCharm now provides native Celery support with a dedicated run/debug configuration that automatically reads your Django project settings. This integration enables direct debugging of asynchronous tasks and includes code completion for Celery jobs and configuration files.

Databases

• AI chat integration for databases - The AI chat integration for Codex and Claude Agent now offers full, native support for your connected databases. With that, you can now query, analyze, and modify your database state using natural language right from the IDE. The same functionality is available for external agents via an MCP server.
• Data source templates in JetBrains Account - Data source settings can now be stored in your JetBrains Account via data source templates. Especially nifty for All Products Pack users or anyone who uses multiple instances of JetBrains IDEs, this upgrade allows you to access data source templates and settings in every JetBrains IDE with database functionality.

• Code Insight
  • Don't navigate to skeleton file when pyi is available.
  • Function return type inlay hints have a trailing space.
  • ty.exe unexpectedly returns an error when selected from /.venv/Scripts.
  • java.lang.IllegalArgumentException in TypeEvalContext in PyCharm 2025.3.0.
• Docker
  • Can't add Virtual Interpreter with Docker Compose.
• Interpreters Settings
  • "Select python interpreter" should support selecting the venv folder.
  • Impossible to create a poetry interpreter with a lower version of Python.
• Jupyter
  • Jupyter Structure View: execution statuses remain after kernel shutdown.
  • Environment variables are not available on Jupyter notebooks when running via IDE Management Server.
  • Drag'n'Drop of a csv file into a Jupyter notebook does not work.
  • Problem with an opening project caused by Jupiter plugin.
• Jupyter: Editor
  • New update broke multiple carets In IPynbs.
• Jupyter: Outputs
  • Plot isn't shown while d2l model is training.
• Packaging
  • Tool installation is not recognized in settings page when added from cmd.
  • pyproject.toml does not reflect the environment state when a package is added from cmd.
• Usability
  • Python SDK in module/.venv does not refresh.
• Tables (Jupyter, DataView)
  • Column Statistics is not displayed for tables with pandas 3.0.0.
  • PyDataViewer.apply: Numpy is not available.
• No subsystem
  • LSP: IDE doesn't show usages when the 'ty' tool is enabled in 2025.3 and using "Declaration or Usages" action.

• Jupyter Notebook doesn't show output of %run after PyCharm updated to 2025.3.2.

• PyCharm 2025.3.2 adds local IDE workflows and Colab-hosted notebooks together. Google Colab support is now available for in PyCharm as a core feature, along with basic Jupyter notebook support. If you already use Google Colab, you can now bring your notebooks into PyCharm and work with them using IDE features designed for larger projects and longer development sessions.

• Refactoring
  • "Invert 'if' condition" refactoring produces never-matching check.
• Build. Maven
  • Memory leak at MavenProjectsTree.
  • Freezes when opening big Maven project (at org.jetbrains.idea.maven.project.MavenProject$Companion.read).
  • High memory usage and freezing opening large Maven projects in 2025.2.6+ (at MavenProjectsManager#initProjectsTree).
• Core. Eel
  • WSL project settings are reset once after updating from 2025.3 or earlier versions.
• Core. WSL
  • WSL2 project settings are reset after restart.