* feat(settings): HTTP Proxy settings for the IDE's outgoing connections
Add Appearance & Behavior › System Settings › HTTP Proxy with three modes:
No proxy, Auto-detect proxy settings (the default; environment proxy as
before) and Manual proxy configuration (HTTP or SOCKS, host and port, a
comma-separated "No proxy for" list with `*` wildcards, optional proxy
authentication). "Check connection…" GETs a URL with the settings as edited
and reports the status or the error.
- ide-core: `ProxySettings` plus routing (no-proxy globs, loopback always
direct, credentials percent-encoded), the current proxy for the process,
curl arguments and child-process variables.
- ide-http: `client::build` / `client::shared` is the one HTTP client
constructor; the HTTP Client requests and the update check go through it.
- ide-tools: curl downloads (tools, JDKs, debug adapters, agent registry,
remote server) follow the setting.
- A manual proxy reaches child processes (language servers, debug adapters,
terminals, git) as HTTP_PROXY / HTTPS_PROXY / ALL_PROXY / NO_PROXY.
- The proxy password is kept in the owner-only saved passwords file
(`proxy-password.json` in the app data directory), not in settings.toml.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(host): Allow the Unix-only post-start test helpers on Windows
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
The remote server read the login shell's PATH once and cached it, had no project-level environment, and changed environment variables only took effect after reinstalling or a manual restart.
- Every openidea-server start re-reads the remote login shell's full environment (`$SHELL -ilc env -0`), with the cached PATH only as a fallback when the shell doesn't answer.
- A project's `.devcontainer/devcontainer.json` (or `.devcontainer.json`) applies on remote hosts: `remoteEnv` (with `${localEnv|containerEnv|env:X[:default]}`, workspace variables and `null` to unset) reaches every child process through one environment overlay (terminals, run configurations, language servers, debuggers, build tools, git), and `postStartCommand` (string, array or parallel object) runs once per server start in the project root as a cancellable background task; a failure raises a notification with the exit code and the last output lines without blocking the project.
- Reload Environment (`remote.reloadEnvironment`, File menu and Find Action) re-resolves the login shell and `remoteEnv` without restarting the server, starting from the process's original environment so prepended PATH entries don't pile up; it lists the changed variables, notes that running processes keep their environment, and offers to restart language servers.
- The JSONC reader moves to `ide_core::jsonc` and is shared with ide-run.
Claude-Session: https://claude.ai/code/session_01Dq6jQQn8ZhRfkkyj3PRutA
Background tasks often couldn't be cancelled, or cancelling didn't stop them:
- A cancel that arrived before the backend registered the request was dropped (build while saving files, multi-repository fetch/pull between repositories). The host now remembers early cancels (bounded, 2 min), and request ids start at a random value per page.
- There was no queue and no common cancel path: backend-reported tasks (LSP progress, deployment transfers) couldn't be cancelled, and tool installs, VCS operations and update checks had no cancel button in Background Tasks.
- Only the direct child process was killed, and JDK / tool archives were extracted without checking for cancellation.
Tool installs and JDK downloads now share a downloads queue; cancelling a queued task removes it without starting any work. Every task cancels through one path that shows "Cancelling…" and then removes it; tasks that can't be interrupted show a disabled cancel button with the reason. IdeEvent::Task carries `cancellable` with a cancel_task command, cancellable LSP progress sends window/workDoneProgress/cancel, installers, curl and tar run in their own process group and are killed as a group, and extraction checks for cancellation between entries.
Claude-Session: https://claude.ai/code/session_01Dq6jQQn8ZhRfkkyj3PRutA
Agents can run through native drivers or any ACP agent, speak ACP v1 or v2, and be installed from the ACP Registry. Tasks gain fork, in-place edit-and-rerun, steering and usage-limit handling where the agent supports them.
- Native Claude Code (`claude` stream-json) and Codex (`codex app-server`) drivers, each served to the task machinery as an in-process ACP shim, using the newest install on PATH.
- ACP v2 with v1 fallback, negotiated on initialize; per-agent capabilities (fork, rollback, steer) so the UI only shows supported actions; usage-limit errors put the task in a `limited` status with the reset time.
- Settings › Tools › Agents lists Registry agents (featured first) to install, update and remove; installed agents update when a project opens. "More agents…" in the provider picker opens it. Registry index and downloads are HTTPS-only, release archives are checked against their sha256, and entries without one show "No checksum provided" and ask for confirmation before installing. Package names from the registry can't be passed as options to npm/uv.
Claude-Session: https://claude.ai/code/session_01Dq6jQQn8ZhRfkkyj3PRutA
Installing an update gave no feedback: app_install_update discarded the download progress callbacks and downloaded the whole package silently before restarting, the UI had no pending state, and repeated clicks started parallel downloads. Manual Check for Updates also showed nothing while checking.
The update now runs as a status-bar background task: "Checking for updates…", then "Downloading OpenIDEA x.y.z" with size and percentage (streamed over a Channel, one event per percent) and Cancel until the download completes, then "Installing…". A running check or install is shared, so clicks from the notification, the notification log, Help › Check for Updates, the toolbar menu, the welcome screen or Search Everywhere join it; the backend allows one install at a time. Failures show an error notification with Retry. After installing, the app restarts through every window's unsaved-changes check and leaves an "installed" notification with Restart in case a window cancels it. Signature verification, endpoints and updater configuration are unchanged.
Fixes#156
Claude-Session: https://claude.ai/code/session_01Dq6jQQn8ZhRfkkyj3PRutA
Adds three split actions to the editor tab menu, the Window › Editor Tabs menu and
Find Action: Open in Opposite Group keeps the tab and opens a copy in the
neighbouring group (splitting when there is none), Change Splitter Orientation
flips the split holding the current group, and Unsplit All closes every other
group.
View > Context Info (Ctrl+Shift+Q on macOS, Alt+Q on Windows) shows the
class, method and block header lines around the caret in a hint, which helps
when they are scrolled out of view.
Files opened from inside the app (File › Open File…, a drop on the Welcome screen, Open File in a single-file window) were always kept in the requesting window, so a file belonging to a project open in another window went to the single-file window instead. These paths now follow the same rule as Finder, Dock and second-instance opens: window_open_files normalises the paths and routes them through plan_open via plan_open_files, which sends files inside an open local project to that project's window. It returns only the files the requesting window should open itself. The TypeScript side no longer has any routing logic. Files dropped onto a project window still open in that window.
Refs #128
Editing and folding
- Move Caret to Code Block Start / End with Selection, Auto-Indent Lines,
Move Element Left / Right (swaps call arguments, parameters and list
entries).
- Fold Code Block folds the {...} block around the caret; Expand to Level
and Expand All to Level 1-5.
- Custom folding regions (//region, #region, <editor-fold>) fold to their
description; Navigate > Custom Folding lists them.
- Error Description shows the problems at the caret; Next / Previous
Highlighted Usage steps through usage highlights.
- Editor tooltips stay below the editor's top edge, so hints on the first
lines no longer hide under the tab bar.
Navigation, project and run
- Navigate > File Path lists the file and its folders to reveal in the file
manager.
- Copy... (F5) opens the Copy dialog for the selection or the open file; the
project view's Refactor submenu gets Copy File.
- Debug... chooses a configuration to debug; Quick Evaluate Expression shows
the selection's value in an expandable hint.
- Inspect Current File runs the inspection without the scope dialog.
- Surround with Live Template lists only $SELECTION$ templates; Java gets
Callable, read / write lock and iterate templates.
Windows
- Stretch the active tool window left / right / up / down, Restore Default
Layout, Go to Next / Previous Splitter, Close Active Tab on a second key,
Stop Background Processes.
All actions have default shortcuts in both keymaps, VS Code preset entries,
main menu items and Chinese translations.
Project view and Keymap settings
- Project view Options: Sort by Name / Type / Modification Time (also in
Find Action).
- Settings > Keymap: Add Mouse Shortcut records a side-button click with
modifiers and checks conflicts like keyboard shortcuts.
- Folding: Expand to Level / Expand All to Level 1-5.
Run configuration detection only looked for pom.xml / build.gradle(.kts) at the opened root, so a Maven or Gradle build in a subfolder produced no configurations, and Java files fell back to a single-file `java X.java` launch without classpath or build. Detection now takes the root build, or else the shallowest build on each branch up to three levels down, with the subfolder in the config's cwd, id and name; a multi-module parent pom counts as one build. Spring Boot applications (@SpringBootApplication with main in a module that applies the Spring Boot plugin) get spring-boot configurations that build the module and run with its classpath. Gutter and file runs pick the build the file belongs to (also for Go modules), and only files outside any build run as single files; the main config lookup takes the nearest bootRun / run / spring-boot:run.
On Windows, piped JVM (java, Maven, Gradle) and Python processes now write in the console encoding instead of the ANSI code page, so their output is no longer garbled. run_configs runs on a blocking thread.
Fixes#149
Claude-Session: https://claude.ai/code/session_01Dq6jQQn8ZhRfkkyj3PRutA
On macOS, OpenIDEA is now listed under Finder's Open With for text and source files and for folders. It never becomes the default app for any of them: all document types are declared with rank Alternate, and only in tauri.macos.conf.json, so Windows and Linux installers register nothing.
Files opened from Finder, the Dock or a second instance share one routing path (plan_open). A file inside a local project that is already open, or still opening, goes to that project's window, choosing the deepest root when projects are nested. Any other file goes to the existing single-file window, or to a new one. A folder focuses the window that already has it open, otherwise it opens as a project. Paths that arrive before setup are queued and opened with the first launch.
The single-file window now closes with its last tab, restores its files the next time it opens, and only offers Open File and drop hints when empty. Also fixes a regression: reopening projects on startup no longer replaces a file opened at cold start.
Refs #128
Files can now be opened without a project. A file dropped on the Welcome screen, picked with File › Open File…, or passed on the command line or to a second instance opens in a lightweight window: editor tabs only, with no project tree, indexing, language servers, VCS or project tool windows. Syntax highlighting, find, save, undo and Local History still work. More files opened later join the same window as new tabs. Open as Project in the status bar opens the file's directory as a full project, following the open-project policy, and reopens the file there.
Backend: a light session reuses Project with an unindexed index, no LSP or VCS, and a flat watch on each opened file's directory. It is never reused as a project for the same root. Previously a file passed at launch opened its parent directory as a project.
- Settings › Commit › "Use non-modal commit interface": when off, Local
Changes, Shelf and Stash become tabs of the Git tool window and ⌘K opens
a Commit dialog; switches live.
- Git tool window: Worktrees tab; Show History and Show Git Log for a
branch open their own closable tabs instead of changing the Log filters.
- Log: toolbar and context menu for a commit's changed files, vertical
toolbar, Tags node and hiding for the branches pane, Collapse / Expand
Linear Branches as separate actions, and a Version Control › Log page.
- Branch popup: collapsible groups, branches grouped by prefix, an options
menu behind the gear, incoming / outgoing icons and "Pull into … Using
Rebase / Merge"; New Branch on ⌥⌘N outside the editor.
- Git main menu regrouped with Uncommitted Changes and Current File
submenus; rebase / merge controls enabled only while one is in progress.
- Settings: Version Control lists its pages; new options for clearing the
initial commit message, showing Update options, confirming dropped
commits, revealing created patches, large-file and file-name warnings,
the Push dialog after Commit and Push, combined stashes and shelves, and
removing applied files from the shelf.
Run oxfmt once over the whole repository (formatting only, no logic changes) and add `pnpm format:check` to the frontend CI job so unformatted code no longer builds up. `.oxfmtrc.json` now skips the generated theme docs, test fixtures and code blocks inside Markdown.
Co-authored-by: jaw52 <53564781+jaw52@users.noreply.github.com>
* feat(agents): Show subagents as their own conversations
Agents that support ACP subagent sessions (Claude Code, Codex) now run each
subagent as a child session. Its tool calls and messages no longer mix into
the task's conversation: the parent shows a card per subagent with its state,
tool call count, latest step, task and result, and the card opens the
subagent's own conversation.
- Route session/update by session id; subagent settings and usage stay out
of the task.
- Subagent permission requests are answered in the task's conversation,
named after the subagent.
- Subagent conversations are saved with the task; a running subagent is
marked stopped after a restart.
- A running subagent can be stopped on its own.
* fix(agents): Save a subagent's state when it changes after the turn
* fix(app): Survive panics in background work and log them
A panic in background work (a spell check, an LSP handler) aborted the whole
app and left nothing behind: the message went to stderr, which a launched app
doesn't keep, and the stripped binary gave crash reports no function names.
- Release builds unwind, so a panic fails one task or command instead of
quitting the IDE. Commands already turn a panicked blocking task into an
error, and the existing catch_unwind around JSON-RPC notification handlers
now takes effect.
- A panic hook appends the thread, location, message and backtrace to the
log file synchronously, before the non-blocking writer could lose them.
- Release builds keep the symbol table (strip = "debuginfo"), so crash
reports and logged backtraces name functions.
* test(app): Match the panic location with file!() so the test passes on Windows
- The title bar project popup lists Remote Development… next to Open… and Clone Repository….
- Opening a remote project follows "Open project in" (new window / this window / ask) when this window already has a project, and confirms leaving the current one first.
- A remote project already open in another window focuses that window.
- A new window connects to the host itself and shows the connection progress on its welcome screen.
- Recent remote projects get Open in New Window.
Replace the mark, app icon, wordmarks and splash logo with the new
design: an orange-framed dark tile with a monospace "oi" on a light
rounded backdrop. Platform icons (png/icns/ico) are regenerated from
src-tauri/icons/source/openidea.svg.
* feat(editor): Drag tabs and files between split panes
HTML5 drag and drop never reaches the webview while native file drop is
enabled, so tab, project-file and changelist drags did nothing. They now
run on pointer events:
- Drag an editor tab onto another pane, or onto a pane's edge to split it.
- Drag files from the Project view into an editor pane (center opens them
there, an edge opens them in a new split).
- Drag files between changelists in the Commit view.
- "Drag-and-drop with Alt pressed only" applies to tabs and stripe buttons.
* fix(lsp): Deliver server settings and start servers on a supported JDK
- workspace/configuration answers a server's own section (basedpyright)
from its settings block; basedpyright runs in "standard" mode and keeps
auto search paths on, so `src/` layouts resolve from `tests/`.
- jdtls: formatting follows the editor indent, calls complete without
guessed arguments, and JDK internals stay out of completion.
- `max-java` in a language server config picks an installed JDK no newer
than that (kotlin-language-server fails on JDK 25).
- Server initialization may take up to 5 minutes (first Gradle import).
- `lsp_resend_diagnostics` re-emits stored diagnostics after a reload.
- Spelling only checks declared names (not called methods like
`substring`) and accepts plurals of known software terms.
* feat(editor): Quick fixes, completion and navigation while typing
Quick fixes
- Errors get a red quick-fix bulb; Apply Quick Fix (⌥⇧↵) fixes the most
severe problem at the caret, and the error hover offers the fix directly.
- Code generated by a code action is re-indented to the file's style.
- Python: Implement abstract methods (⌃I or the bulb) from the bases.
- Duplicate code actions are listed once.
Completion
- Overloads sharing a name, and a name importable from several modules,
are no longer merged into one entry.
- Auto-import entries show the module they import from.
- Call snippets without a stop (`new ArrayList<>()`) leave the caret inside
the parentheses with parameter info, or past them when nothing takes
arguments; parameter info also opens after server call snippets.
- The selected item is resolved ahead of acceptance, so its auto-import
is applied (rust-analyzer).
Navigation and diagnostics
- Language server requests wait for pending edits, so Go to Declaration
right after typing answers for the current text; an empty answer right
after an edit is retried while the server catches up.
- Diagnostics come back after a window reload; error underlines stay
visible under typo / unused highlighting.
- Implementation gutter markers no longer refresh in a loop while the
server reports progress.
- A missing alternative server no longer shows as a server error.
* chore(lint): Skip agent worktrees under .claude
* fix(editor): Skip cancelled LSP requests and limit generated-code reformat
- Requests cancelled while waiting for pending edits are not sent.
- Code generated by an action is reformatted per changed hunk, so an added
import doesn't reformat the code between it and the generated lines.
- A long busy spell caused by a settle refresh re-runs it only once.
- Files dropped on a file row of another changelist move to that changelist.
- Python Implement methods: a nested class without bases ends the lookup,
class bodies continue through unindented string lines, and
`@abstractproperty` stubs become `@property`.
- Resend diagnostics goes through the host ops, and custom servers get
`max_java`.
* fix(editor): Rank the exact type first in Java constructor completion
jdtls appends the call or body to filterText (`String(${0})`); matching
on the label instead lets typing `String` rank `String` above
`StringBuilder` and the other types sharing its sort text.
* fix(editor): List identical completion items once
kotlin-language-server repeats a class once per classpath entry declaring
it, so `String()` and `String` were listed three times each.
* fix(editor): Keep the internal cancellation message out of the UI text catalog
Language servers can be configured from settings.toml (global or project) and Settings › Language Servers: arguments, environment, initialization options and workspace settings per server, enable/disable, and any stdio LSP server can be added for a language. Settings-only changes are pushed with workspace/didChangeConfiguration; launch changes restart the server.
Also: re-enabling a server now detaches and stops the fallback that stood in for it, and a server added from the dialog is written as its own [languageServers.<name>] table.
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Adds gutter markers for type hierarchies: ↓ "Is implemented / Has subclasses / Is overridden" on types and methods with implementations, ↑ "Implements / Overrides method in X" on methods that implement or override a supertype method. One target navigates, several open the Choose Implementation popup. Settings › Editor › General › Gutter Icons gets two toggles.
Markers come from the language server: a new batched `lsp_implementations` command (up to 4 concurrent, cancellable, sharing the goto query path) for ↓, and the Super Method lookup for ↑, which now prefers the overload with matching parameters (also fixes ⌘U on overloads). Only declarations near the viewport are queried, debounced, cached per document version and cancelled when superseded. Works for any server with implementation / type hierarchy support; servers without type hierarchy show ↓ only.
Fixes#99
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
The new-task composer only offered the slash commands the agent sent in its last session, stored once globally, so a project that hadn't run a session (or whose last session ran elsewhere) showed none of its skills; matching was substring-only, so `/trig` didn't find `/triage` either.
Offered commands are now remembered per project root. For Claude Code, skills (`.claude/skills/*/SKILL.md`) and commands (`.claude/commands/**/*.md`) are also discovered on disk, from the working directory up to home plus `~/.claude`, and merged with the agent's list (agent entries win). New `agent_commands` IPC serves the home composer (project root) and session composer (the task's cwd). Matching ranks prefix (also after `plugin:`), then substring, then subsequence. Picking a completion inserts `/name ` and the prompt is sent unchanged.
Not covered: a new worktree of a project that gitignores `.claude/` doesn't contain the project's skills.
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
On Windows the auto-downloaded jdtls ran through `bin\jdtls.bat`, which starts the Python wrapper and ends with `pause`. When Python failed, `pause` wrote its prompt to stdout in the console code page; the transport treated that as a broken LSP header ("non-UTF-8 header"), dropped it and the server was killed, so the status only said "exited with code 1".
- jdtls is now launched with `java` directly on every platform (Equinox launcher, platform config dir, JDK 24+ flags, forwarded JVM args, `-data`), so Python is no longer needed; user wrappers and unusual layouts keep their command.
- Output outside the protocol on stdout (and non-UTF-8 stderr) is decoded (UTF-8, then the Windows OEM/ANSI code page, then lossy), logged and included in the crash message, for stdio and TCP language servers, DAP adapters and ACP agents.
- Crash messages carry the real exit code and the last lines of output.
未在 Windows 实机验证 (Windows paths covered by unit tests and a Windows-target clippy run).
Fixes#90
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Adds Solarized Light, Solarized Dark, One Light and One Dark as built-in themes across the workbench, editor, terminal and Mermaid, keeping the original palettes (MIT notices in THIRD_PARTY_LICENSES.md, mapping in docs/themes.md).
- Native window colours are read from tokens.css at build time (`theme_frame.rs`) instead of a second colour table.
- Startup restores the built-in palette before first paint.
- Editor schemes, the terminal and Mermaid refresh on switches between themes of the same kind.
- The plugin packages of the same themes are removed.
- A test requires every built-in palette theme to define each token (only listed ones may inherit from the base theme), so new tokens can't silently miss them.
Co-authored-by: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Maven sync ran a plain `dependency:tree`, which drops the nodes Maven omits (conflict losers, duplicates), so there was no data to show where a conflict comes from.
Sync now runs `maven-dependency-plugin:3.8.1:tree -Dverbose` (the project's own plugin version when offline) and records omitted / conflictWith / managedFrom per node; omitted nodes stay out of the Maven tool window and External Libraries, and older caches are marked stale. The new Dependency Analyzer (Maven toolbar, module / Dependencies node menus, pom.xml menus, Find Action) lists a module's resolved artifacts or tree with conflict markers, search and scope filters, and every path to the selected artifact with its status. Exclude writes an `<exclusion>` into the declaring pom (module or in-project parent), keeping indentation and line endings, then reloads; Jump to Source and Copy Coordinates are on each path. Gradle is not covered yet.
Fixes#87
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Solarized Light/Dark and One Light/Dark ship as theme plugins (plugins/theme-solarized, plugins/theme-one) with complete UI tokens, editor schemes and terminal colours, and the original palettes' MIT notices. Installing a plugin now also copies its LICENSE/NOTICE/COPYING.
Theme handling that benefits every plugin theme:
- The last plugin theme is restored before first paint (splash, workbench); the native window background follows the theme.
- HSL / color-mix() / var() colours are resolved to RGB for native windows, the Settings window and Mermaid.
- Mermaid, the terminal and editor schemes follow switches between two themes of the same kind.
- An explicitly chosen editor scheme wins over a plugin theme's editor colours.
Supersedes #85.
Co-authored-by: bghtyu <3931591+bghtyu@users.noreply.github.com>
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Reference-based features only asked the language server, which analyses its configured scope (e.g. Pyright's `include`) plus open files, so usages in unopened files outside that scope were silently missing.
Find Usages, Show Usages, usage counts, Safe Delete of a symbol, Rename and incoming calls now also search project files that contain the symbol's name (whole word, same language, .gitignore/excludes respected, unsaved edits included, up to 300 files, cancellable). Candidates the first answer didn't cover are opened in the server temporarily, queried and closed again. Files the user has open are never touched, and diagnostics for the temporary files are dropped. An empty result while the server is still indexing is retried once indexing ends. When results may still be incomplete (indexing, file cap), the usages UI says so. `lsp_references` takes a RequestId and is cancelled by newer searches. Not covered yet: the Safe Delete scan when deleting a file or directory.
Fixes#83
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
An app started from Finder or the Dock took only PATH from the login shell, so
variables exported in rc files never reached agents, language servers, run
configurations or debug targets. Claude Code configured for Bedrock through
CLAUDE_CODE_USE_BEDROCK / AWS_REGION in ~/.zshrc fell back to the Anthropic API
and failed with 403.
The shell's environment is read once at startup (in the background when PATH
comes from the cache) and kept in memory only, since it may hold secrets. Spawns
add the variables this process lacks before their own.
In projects with several repositories the Branches popup began with a wrapping row of repository chips and listed only the selected repository's branches, and the widget always read "repo / branch".
The popup now lists one row per repository (colour, name, current branch) whose submenu holds that repository's branches. When all repositories are on the same branch the widget shows just the branch and the popup offers Common Local / Remote Branches whose actions apply to every repository; otherwise it shows the current repository's Recent / Local / Remote Branches and Tags. The widget and the popup use the same current repository: the one owning the active file, else the selected one. Search covers actions, repositories and branches across repositories. New Branch… creates the branch in all repositories. Single-repository projects are unchanged. `vcs_roots` now returns each repository's branch and HEAD.
Fixes#78
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Adds GDScript, Godot resource and shader file types with tree-sitter highlighting, the GDScript language server over TCP (a headless Godot editor started on demand), "Godot Project" / current-scene run configurations, and debugging through Godot's debug adapter.
Breakpoints set through symlinked paths (/tmp, /var on macOS) were reported verified but never hit; the adapter now sends resolved paths and maps stack frames back to the paths the IDE opened. The duplicate exit event Godot sends after disconnect is ignored.
Fixes#68
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
## What
An Agents feature: hand a task to a coding agent and follow it from the IDE. Agents speak the [Agent Client Protocol](https://agentclientprotocol.com). Built in: Claude Code, Codex, Gemini CLI, GitHub Copilot, Cursor, OpenCode, Qwen Code, Kimi CLI, Goose, Auggie, Kiro, Mistral Vibe. You can add your own in `<config>/agents/<id>.toml` (`name`, `command`, `args`, `[env]`). An agent whose command isn't on `PATH` is listed under "Not installed".
- **New Task page.** An editor page with a prompt box. You pick the agent, model, mode and thinking effort, and whether it runs in a new worktree (from a chosen branch) or in the checkout.
- **Agents tool window.** Tasks are grouped by what they need: Needs Input, Failed, Running, Ready to Review, Archived. It also shows a stripe badge, a status bar item and notifications.
- **Session tab.**
- The conversation: Markdown, thinking, tool calls with diff stats, the plan.
- Permission requests appear as an editor banner.
- The changed files are compared against the task's base commit and open in the diff viewer.
- A context-usage meter.
- **Composer.**
- `@` references files from the project index. A referenced file's text is sent when the agent accepts embedded context, otherwise a `resource_link`.
- `/` lists the agent's commands; ↑/↓ recall earlier messages.
- Images can be pasted, dropped or picked; Shift+Tab cycles the mode.
- While the agent works, Enter queues the message and ⌘↵ interrupts the turn and sends it now. Queued messages can be sent now, edited or removed.
- **Send to Agent.** In the editor and Project view context menus. It adds `@file:10-20` references to the message you typed last.
- Landing the work is left to the existing Git features: the worktree is a normal branch.
## How
- **`crates/ide-agent`**: one ACP agent process per task, running in the task's directory.
- It folds `session/update` into a timeline and answers `session/request_permission`.
- It tracks the session's settings: `configOptions`, falling back to the older `modes` / `models` fields. The user's choices are applied again on every new or reloaded session.
- Tasks are saved as JSON per task. Sessions resume with `session/load` without duplicating history, and old task files are upgraded when read.
- **`ide-jsonrpc`**: adds newline-delimited framing (ACP) next to Content-Length (LSP).
- **`ide-vcs`**: adds `numstat_with_worktree` for the changes list.
- **Editor**:
- The Switcher (⌃⇥) lists every open editor, as IDEA does, not just files.
- `EditorKindContribution.restorable` lets a kind's tabs be saved with the editor session, so agent tabs survive reloads.
- **UI**: `Select` popups now size to fit short option details (they used to cut them off).
## Testing
- `cargo test -p ide-agent` (14 tests). They use an in-memory ACP agent and cover permissions, restart with `session/load`, settings, the queue with send-now and stop, and prompt blocks.
- There is also an ignored end-to-end test against the real Claude Code adapter: `OPENIDEA_AGENT_E2E=1 cargo test -p ide-agent real_claude -- --ignored`.
- `pnpm vitest run`: 2052 tests.
- `cargo clippy -p ide-agent -p openidea --all-targets -- -D warnings` and typecheck are clean.
- Driven in the running app with Claude Code:
- A task in a worktree, with permission answers, changes listed and the diff opened.
- A follow-up after restarting the IDE.
- `@` reference with embedded content, `/` commands, Shift+Tab mode changes.
- A queued message and send-now, an image prompt, Send to Agent from a selection.
- The Switcher and restored tabs after a reload.
Not covered yet: only Claude Code was driven end to end. The other agents go through the same ACP path but weren't run here.
Merged on the maintainer's go-ahead. Merge with main: kept both sides in Cargo.toml (ide-agent + ide-clipboard), the feature registry (agents + vim) and zh.json (no duplicate keys). After the merge: typecheck clean, full vitest 268 files / 2247 tests passed, ide-agent/ide-jsonrpc/ide-vcs 144 passed, bindings unchanged after regeneration.
New query, data and schema tools for the Database feature (SQLite, PostgreSQL, MySQL).
## Query console
- **Explain Plan** tab: SQLite `EXPLAIN QUERY PLAN`, PostgreSQL `EXPLAIN (FORMAT JSON)`, MySQL `EXPLAIN`, parsed in `ide-db::plan` into an operation tree (Full Scan / Index Scan / Group By / Order By …) under a statement root, with a properties panel.
- **Query history**: `ide-db::history` keeps executed statements per project and source in `db-history.json` (deduplicated, newest first, 500 max). The Browse Query History dialog has search, a highlighted preview, Paste and Delete.
- Output reports `(execution: X ms, fetching: Y ms)`.
## Results and table editor
- Local column filter (value list with counts) and local sort.
- Record / Value / Aggregates side panel.
- Chart view (bar / line) with X / Y / aggregation settings.
- Export Data dialog: extractor, Transpose, headers, preview, file or clipboard. An extractor drop-down sits on the data toolbars.
- Import Data from a CSV / TSV file.
- Table editor gets a WHERE | ORDER BY row. Column headers fit their names, with filter and sort indicators. Whole SQLite REALs display as `3108`, not `3108.0`.
## Database tree
- Triggers under tables. Routines, sequences and materialized views under schemas, with DDL, Quick Documentation and Drop.
- Data source options: object filter, show template databases, single database mode (PostgreSQL).
- New ▸ Import from Clipboard (connection URLs).
## IPC
New commands: `db_schema_objects`, `db_object_ddl`, `db_explain`, `db_history`, `db_remove_history`. Bindings are regenerated.
## Testing
- `cargo test -p ide-db`: plan parsing, history store, trigger listing and DDL, explain without executing.
- `cargo clippy -p ide-db -p openidea --all-targets -- -D warnings` is clean.
- `pnpm typecheck` and eslint pass. `pnpm vitest run src/features/database src/ui`: 226 passed.
- Checked in the running app against a SQLite database:
- tree with triggers
- result grid, local filter, side panel
- chart
- Explain Plan
- history dialog
- Export Data dialog
- table editor ORDER BY
- CSV import
Merge follow-up: grouped the new `--chart-series` token under a section comment and added it to the theme plugin template (theme.test.ts on main), regenerated docs/plugins/themes.md. Re-verified after merging main on macOS with a SQLite source: query + result grid with filter/sort headers, Explain Plan tab (Full Scan / Index Scan / Group By / Order By), Query History dialog; `cargo nextest run -p ide-db` 45 passed, bindings unchanged after regeneration.
https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Root cause: the IDE only detected installed JDKs and had no way to get one, so any launch needing a newer JDK than the machine has (latest jdtls needs JDK 25) dead-ended in an error.
Fix: IDEA-style Download JDK (Eclipse Temurin via the Adoptium API, sha256-verified, reusing ide-tools' download/verify code; Windows zip, Linux tar.gz, macOS Contents/Home) into <app-data>/tools/jdks, registered with JDK discovery and rescanned after each download. "Download JDK…" in the Project SDK and Language Server JDK pickers; a one-click `java.downloadJdk` "Download JDK N" on "needs JDK N" errors, which also repoints a too-old language server JDK and restarts jdtls.
Siblings: language server status popup, jdtls launch notification, run console classpath error, debug launch, run configurations without a JDK ≥ the module's Java version, Maven/Gradle sync failures, single-file Java run without `java` on PATH.
Verified in the app on macOS: vendor/version/location queries, the Download JDK dialog from the Language Server JDK picker, a real Temurin 25.0.5 download that runs. Not verified on Windows/Linux.
Fixes#49https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Built-in Vim emulation on @replit/codemirror-vim (approved dependency), modelled on IdeaVim: `vim.toggle` (Tools › Vim, status widget, Settings › Editor › Vim), mode widget (NORMAL / INSERT / VISUAL / V-LINE / V-BLOCK / REPLACE), IdeaVim-style shortcut conflict handlers (Vim / IDE per key) through a generic `registerKeyClaim()` in the keymap dispatcher, Escape closing popups first, a safe subset of ~/.openideavimrc / ~/.ideavimrc (set, map variants, mapleader, :action / <Action>()), Ex commands :w :wa :q :qa :wq :x :action. Covers file editors, diff editable side and merge result via a new `embeddedEditorExtensions` extension point.
Verified in the app on macOS: motions/operators, i/Esc/V mode widget, :w saves, Ctrl+R redoes in Vim without running Run, toggling off restores plain editing. Not verified on Windows/Linux.
Fixes#27https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Root cause: `uses_lombok` only scanned build files ~3 levels below the jdtls root, so Lombok inherited from outside that tree was judged `Unused` and no `-javaagent` was attached — even when Lombok was installed in Settings.
Covered: parent pom only in the local Maven repository (settings.xml localRepository honoured, multi-hop, cycle-safe), parent via relativePath outside the workspace, outer Gradle root build / version catalog, `lombok.config` in ancestor directories, and a bounded source fallback (`import lombok.` / `@lombok.`) for convention plugins and BOMs. A Lombok jar installed in Settings is now always attached (like IDEA's always-on Lombok support). Detection in `java_resolve_launch` runs in `spawn_blocking`.
Fixes#33https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Root cause: the Git Log already supported multi-selection, but the Changes pane only read the last clicked commit and the backend had no "changes across several commits" query.
Fix (IDEA): ide-vcs Repo::commits_changes merges per-file changes across selected commits (before the first touching commit ↔ after the last, renames chained, add-then-delete dropped); the details pane shows "N files in M commits" with the commit list; exactly two commits get "Compare Versions" (version ↔ version dialog, also fixing added/deleted sides).
Siblings covered: Show History for file/folder (same log), Local History multi-select (oldest ↔ newest diff).
Verified: ide-vcs tests incl. temp-repo commits_changes, vcs vitest (143), in-app on macOS: 3 contiguous commits → "2 files in 3 commits"; Cmd-click 2 commits → Compare Versions lists the right files.
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Root causes: Typo "Change to" replaced one occurrence even for declared names (breaking references); inspections had no Disable / Edit settings controls anywhere; en_US Hunspell does not derive agent nouns like "checkpointer" and compound identifier words were flagged; local inspections (Typo, file references, SQL) never reached Problems.
Fix (IDEA): "Typo: Rename to…" renames via LSP rename, falling back to syntax-based local rename, never a silent single replace, and lists identifier-safe words only; intention submenu (→) with Edit inspection profile setting / Disable inspection for every inspection; Problems and Inspection Results get Disable Inspection, Change Severity, Edit Inspection Profile Setting; Typo options Process code / literals / comments; -er/-ers derivation and identifier compounds accepted, common misspellings still flagged.
Siblings covered: all LSP and built-in inspections' intentions, file reference and SQL inspections in Problems, severity counter.
Verified: ide-spell tests (issue examples + ~50 real misspellings), vitest (286), in-app on macOS: only 'mesage' flagged, submenu, Rename to → 'message' updated both occurrences.
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Exchange files with the OS clipboard on Windows: CF_HDROP read/write in a new ide-clipboard crate (Win32 FFI behind cfg(windows), stub elsewhere), three thin commands, and project-view copy/paste hooks. Pasting from Explorer always copies; our own hand-off falls back to the internal clipboard so Copy/Move dialogs and refactoring stay intact.
Verified: ide-clipboard round-trip tests, project-tree vitest (93), CI on all three OSes. Not verified on Windows hardware here.
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
Root cause: run configurations could only launch a process; there was no "attach to a debugger that is already listening" type, and debugServer only connected to 127.0.0.1.
Fix (IDEA / GoLand / PyCharm): debug-only Remote JVM Debug (attach / listen, live JDWP agent args per JDK version), Go Remote (dlv --headless, substitutePath), Python Remote Debug (debugpy --listen, pathMappings) and Attach to Node.js (inspector, localRoot/remoteRoot); debugServer accepts host:port and [v6]:port; JVM listen mode via a cancellable relay so java-debug can attach.
Verified: ide-dap tests (host:port, relay handshake, cancel), run/debugger vitest, in-app on macOS: Python Remote Debug attached to a debugpy server, paused at app.py:3 with variables. Remote JVM/Go/Node not exercised in-app.
Claude-Session: https://claude.ai/code/session_01VUYsdXA2nkF5v6zsC93dpn
l0ng-aiandClaude
Jaw
Walter WangandClaude
kevinh