TheiaIDE: versioni precedenti da scaricare

- [ai] support for server-side compaction #17746
- [ai-anthropic] added configurable server-side compaction thresholds #17795
- [ai-anthropic] enabled server-side compaction for dateless major model ids #17814
- [ai-anthropic] surfaced deferred-tool search as a server tool call #17753
- [ai-chat] enhanced command parsing to support multiple slash commands and command disambiguation #17661
- [ai-chat] resolved hidden pinned agents in delegated sessions #17850 - Contributed on behalf of K2view
- [ai-chat-ui] added per-session model selector and in-thread badge #17786
- [ai-chat-ui] blocked external resources in ai chat markdown #17500
- [ai-chat-ui] introduced a consolidated session state #17764
- [ai-chat-ui] relied on virtuoso's onAtBottomStateChange for scroll locking #17728
- [ai-chat, ai-chat-ui, ai-ide] rendered delegated sessions as children in session list #17777
- [ai-core] added AiConfigurationService for AI preferences #17768
- [ai-core] made custom-agent migration opt-in #17744
- [ai-ide] fixed an issue preventing searchInWorkspace tool from hanging #17774
- [ai-openai] added native web search support #17782
- [ai-openai] removed strict schema conversion for the OpenAI Response API #17754
- [ai-openai] updated model defaults #17817
- [ai-openai] updated openai-sdk to latest version to not throw on new heartbeat event #17802
- [ai-registry] fixed correct search query for mcp "from registry" link #17740
- [ai-registry] routed registry fetch through the backend #17736
- [ai-tool-sketchpad] introduced AI Tool Sketchpad for runtime tool prototyping in custom agents #17225
- [ai-vercel-ai] deprecated the @theia/ai-vercel-ai package #17781 - Contributed on behalf of STMicroelectronics
- [core] fixed always show the bottom-panel toggle #17749
- [core] fixed an issue with logger promise always resolving #17730 - contributed on behalf of STMicroelectronics
- [core] fixed browser-only to fail fast instead of hanging when backend services are unavailable #17794
- [core] fixed color list.activeSelectionBackground in light theme #17778
- [core] fixed issue with dead keys in key mappings for glyphs #17683
- [core] fixed preference values merging to ignore reserved keys #17845
- [core] fixed window control style on electron custom title bars #17835
- [core] implemented perspectives + AI First perspective #17832
- [core] improved SelectComponent keyboard navigation #17785 - Contributed on behalf of Texas Instruments
- [core] improved composition check for keybindings #17723
- [core] npm upgrade, bumped dependencies to reduce vulnerabilities and added lockfile maintenance guide #17741 - Contributed on behalf of STMicroelectronics
- [core] required connection token on file endpoints #17806
- [core] Theia v1.73.0 was released #17716
- [core] vscode API bumped and nls updated to 1.130.0 #17819 - contributed on behalf of STMicroelectronics
- [core, ai-ide, ai-chat-ui, preferences, plugin-ext, monaco] added session-preference CLI overrides and AI Allow-All Mode banner #17731
- [doc] recommended interface + symbol over class tokens for services #17792
- [doc] updated developing.md #17752 - Contributed on behalf of STMicroelectronics
- [doc] updated migration.md #17827
- [docs] updated SECURITY.md to Eclipse security-handbook template #17803
- [electron] updated to electron 42 #17586
- [filesystem] fixed file dialog location input navigation #17665 - contributed on behalf of STMicroelectronics
- [markers] added copy as text, export, and keybinds #17717
- [mini-browser, plugin-ext, filesystem, application-manager] fixed some node 24 deprecation warnings #17757
- [navigator] fixed keeping the in-app file clipboard in sync and awaiting pastes #17837
- [navigator] fixed matching files.exclude relative to workspace #17839
- [navigator] fixed support for pasting files from the menu in browser apps #17829
- [plugin-ext] added tabinfo for customeditorwidget and viewcolumn from the tab bar's index #17807
- [plugin-ext] deferred plugin view initializa

- [ai] continued ai white label #17447
- [ai] merged consecutive assistant messages to comply with strict role alternation #17425
- [ai-anthropic] removed old anthropic models and updated gemining flash #17524
- [ai-chat] drove token usage indicator from the active model's context window #17443
- [ai-chat] fixed synthesize result for unfinished tool calls on restore #17496
- [ai-chat] improved chat session card UI and provider error presentation in chat UI #17502
- [ai-chat] made tool confirmation default a dedicated preference #17452
- [ai-chat-ui] fixed chat capability selections resetting on refresh #17480
- [ai-chat-ui] fixed chat header reflow during "generating" animation #17492
- [ai-chat-ui] removed auto-generated groups from generic capabilities functions #17470
- [ai-chat-ui] showed description and arguments in tool confirmation #17439
- [ai-code-completion] fix(ai-code-completion): disable reasoning for inline completion #17521
- [ai-code-completion] fixed unit tests #17527
- [ai-core, ai-mcp] cleaned up tool registrations when MCP down #17407
- [ai-ide] allowed read tools to access user-allowlisted external paths #17419
- [ai-ide] improved pr-reviewer workflow #17532
- [ai-ide] improved user-interaction tool #17530
- [ai-ide] introduced new "fast" model alias for simpler agents #17535
- [ai-openai] fixed openAI responses API token usage aggregation #17481
- [ai-terminal] implemented terminal command AI variable #17312
- [application-manager] fixed issue with calling yargs as factory in esbuild ESM template #17493 - Contributed on behalf of STMicroelectronics
- [application-manager] fixed parse process.argv in esbuild yargs setup #17544 - contributed on behalf of STMicroelectronics
- [bundle-plugin] aligned webpack ripgrep resolution with esbuild #17560
- [bundle-plugin] fixed issue with esbuild source maps so breakpoints bind #17517 - Contributed on behalf of STMicroelectronics
- [bundle-plugin, file-search, search-in-workspace] supported @vscode/ripgrep>=1.18.0 #17483
- [ci] fixed set-milestone and check-new-package workflow for github-script v9 #17463 - Contributed on behalf of STMicroelectronics
- [ci] updated GitHub Actions to Node.js 24 compatible versions #17460 - contributed on behalf of STMicroelectronics
- [core] fixed escape backticks in markdownstringimpl.appendcodeblock #17440
- [core] fixed filesystem to show git decorations on every file when changes exceed the plugin-ext event cap #17508
- [core] fixed issue with unix process termination #17563
- [core] fixed node expansion when double-clicking checkbox #17372
- [core] fixed omission of '+' separator from menu shortcuts on macOS #17423 - contributed on behalf of STMicroelectronics
- [core] fixed selectComponent dropdown z-index issue #17472
- [core] reverted "fix(filesystem): show git decorations on every file when changes exceed the plugin-ext event cap" #17555
- [core] supported bundling Theia with esbuild #14414
- [core] theia v1.71.0 merged #17438
- [core] Translation update for version 1.72.0 #17572
- [core] used MarkdownRendererImpl in HoverService to fix code block rendering in tool call tooltip #17499
- [core] wrapped long strings in hover popovers #17548
- [debug] added missing space between expression name and error/unavailable message #17495
- [debug] allowed the mouse to summon the context menu in editor gutter on macOS #17422 - Contributed on behalf of STMicroelectronics
- [dev-container] implemented overrideCommand for non-compose containers #17474
- [dev-packages] ran npm upgrade after 1.71.x #17446 - Contributed on behalf of STMicroelectronics
- [doc] added double-dashes so options are passed to npm run script #17486
- [doc] suggested using npm upgrade --save in publishing guide #17511
- [editor] improved navigation support in the merge editor #17242
- [ffmpeg] fixed fallback to package.json when electron dist/version is missing #17569
- [filesystem] bind filesystem proxy as singleton #17424
- [filesystem] replaced fixed sleep timeout

- [ai] replaced the per-model thinking-mode toggle with a provider-agnostic reasoning selector and ReasoningSettings abstraction; the chosen level can be persisted per agent via the chat capabilities save action #17363
- [ai-anthropic] added display: 'summarized' to adaptive thinking block #17429
- [ai-chat] added taskContextId parameter to delegateToAgent tool #17244
- [ai-chat] dropped incomplete thinking blocks from llm history #17395
- [ai-chat] fixed agent label casing in delegation renderer #17234
- [ai-chat-ui] added token usage indicator to chat view #17028
- [ai-chat-ui] fixed inverted expand/collapse in capabilities tree during search #17248
- [ai-chat-ui] fixed issue with preserving user-selected mode during agent delegation #17307
- [ai-chat-ui] integrated tool confirmation system into DelegationToolRenderer #17293
- [ai-chat-ui] show tool confirmation in collapsed delegation summary #16926
- [ai-chat-ui] used partial JSON extraction to prevent tool call label flicker #17306
- [ai-chat-ui] warned when chat session token usage crosses a threshold #17387
- [ai-core] prompt templates do not get modified only by opening them #17277
- [ai-core] provided ready promise for SkillService #17317
- [ai-core] updated default AI model versions for Anthropic and OpenAI #17409
- [ai-core] white label theia and theia ide #17265
- [ai-core, ai-chat-ui, ai-chat, ai-claude-code, ai-code-completion, ai-editor, ai-ide] hooked workspace trust into AI features #17364 - contributed on behalf of STMicroelectronics
- [ai-ide] added PR Review Agent with UserInteractionTool #17298
- [ai-ide] fixed chat session canceling completed tasks #17297
- [ai-ide] improved agent mode confirmation UX and sync mode selector #17271
- [ai-ide] improved code prompt and tool descriptions #17197
- [ai-ide] improved todoWrite tool description #17351
- [ai-ide] provided proper icons for agents and fixed icons in ai config #17336
- [ai-ide] readded junior contributions after release #17259
- [ai-ide] removed junior contributions #17418
- [ai-terminal] added description field to shell execution tool #17350
- [ci] used a single rolling next git tag for npm next publishes #17355 - Contributed on behalf of STMicroelectronics
- [collaboration] fixed resolving of auth endpoint paths for server urls with path #16751
- [core] added .tmp.cfg to top-level gitignore #17377
- [core] added errorHandling option to Emitter to control how listener exceptions are handled: 'log' (default), 'propagate' (collect and re-throw), or a custom callback #17332
- [core] API bump and nls update to 1.116.0 #17367 - Contributed on behalf of STMicroelectronics
- [core] belated breaking changes for #17360 #17410
- [core] fixed accessibility issues in the AI chat view #17260
- [core] fixed copy in chat view not working without selection in Electron #17300 - contributed on behalf of STMicroelectronics
- [core] fixed editor context menu filtering and Monaco widget styling #17404 - contributed on behalf of STMicroelectronics
- [core] improved electron window close behavior on windows #17296 - contributed on behalf of STMicroelectronics
- [core] improved start-up performance logging #17368 - contributed on behalf of STMicroelectronics
- [core] limited imports of code in the preload section #17281
- [core] made socketWriteBuffer and reconnectableSocketChannel overridable via DI #17249
- [core] npm upgrade #17282 - contributed on behalf of STMicroelectronics
- [core] onSaveChange: saved on changes of both tab and window focus #17360
- [core] removed stale beta labels from translation files #17314
- [core] showed empty base for both added conflicts in merge editor #17274
- [core] showed Linux env variable hint in API key preferences #17337
- [core] translation update for version 1.71.0 #17433
- [core] unbound all services in the backend in connection-scoped child containers on close. This allows clean-up methods annotated with @preDestroy() to run, releasing resources. Downstream applications should be

- [ai] added proxy support to AI providers #17085
- [ai-chat] fixed forward change events for restored chat request models #17171
- [ai-chat] fixed undo/redo history preservation when applying AI suggestions #17044
- [ai-chat] multiple chat fixes #17081
- [ai-chat-ui] enhanced the UI and UX of the ClaudeCode AskUserQuestion tool #17003
- [ai-chat-ui] prevent Chat UI crash when tool call returns a primitive value #17203
- [ai-chat-ui] sorted capabilities items by name #17092
- [ai-chat-ui, ai-core, terminal, terminal-manager] added chat & terminal to default workbench & adapted custom agents default model #17177
- [ai-copilot] added model auto-discovery and migrated to OAuth App #17198
- [ai-core] improved the skill slash commands #17182
- [ai-core] persisted capability settings to preferences #17109
- [ai-core] updated default models for 1.70 #17180
- [ai-google] fixed token usage showing incorrect values for gemini models #17173
- [ai-ide] added first-use confirmation for agent mode and made it the default #17209
- [ai-ide] added size limit to getFileContent #17130
- [ai-ide] Alphabetically sorted the allowlist and denylist in the ai configuration tools view #17095
- [ai-ide] fixed delegation for capabilites #17159
- [ai-ide] improved task context prompt for plan execution #17206
- [ai-ide] provided multiple agents to allow for full software engineering flow #17125
- [ai-ide] removed beta labels from AI features #17210
- [ai-ide] removed junior contributions for release #17222
- [claude] updated CLAUDE.md file #17093
- [core] added precheck optimization to LoggerSanitizer rules #17139 - contributed on behalf of STMicroelectronics
- [core] added window zoom status bar item and fixed secondary window zoom #17164
- [core] allowed clicking into hover tooltips to select and copy content #17161
- [core] fixed dependency ranges and updated @types/node #17151 - contributed on behalf of STMicroelectronics
- [core] fixed dropdown issue #17152
- [core] fixed preference imports that refer to 'common' #17121
- [core] fixed refresh of tab-bar toolbar items on context key changes #17106
- [core] fixed title bar shift when opening *.md in preview #17138
- [core] hoisted @lumino/widgets and @lumino/dragdrop to root node_modules #17116
- [core] implemented files.autoSaveWhenNoErrors #17089
- [core] merged changes from release branch #17074
- [core] npm upgrade #16920
- [core] prioritized substring matches over characterwise fuzzy matches #17136
- [core] supported image paste and fixed clipboard in input fields #17189
- [core] translation update for version 1.70.0 #17230
- [core] updated built-in extension versions #17221
- [core] used macOS symbols for menu shortcut labels in browser #17084
- [debug] refactored breakpoint identity model so BreakpointManager is the sole owner of DebugBreakpoint instances, using a VSCode-style sessionData map for multi-session support #17083
- [dev-container] supported for devcontainer workspacemount & workspacefolder #15276
- [filesystem] fixed file dialog path input on Windows #17134 - contributed on behalf of STMicroelectronics
- [filesystem] fixed filesystem issue by normalizing Windows paths for OPFS compatibility #17113
- [git] removed @theia/git extension code entirely (deprecated since v1.58.0) #17148
- [github] improved usage of copilot features and introduced on-demand review agent #17183
- [github] used copilot instructions instead of custom agent #17192
- [github-actions] updated GitHub Actions to Node.js 24 compatible versions #17147 - contributed on behalf of STMicroelectronics
- [license-check] worked around broken dash-licenses download URL temporarily #17169
- [monaco] restricted clipboard/selectAll Monaco handlers to editor text focus #17204
- [monaco] updated @theia/monaco-editor-core to 1.108.201 #17140 - Contributed on behalf of STMicroelectronics
- [monaco, scm] fixed git actions and scm tree selection in diff editors #17158 - contributed on behalf of STMicroelectronics
- [output] fixed 135

- [ai] declared agent-specific variable in claudeCode agent and improved agent specific variable status UX #16967
- [ai-anthropic] added support for custom anthropic models #16673
- [ai-anthropic] updated max tokens for opus 4.6 #17051
- [ai-anthropic, ai-core, ai-google, ai-openai] updated models #17040
- [ai-chat] added mode selector to ask AI input without session overhead or code duplication #16914 - contributed on behalf of EclipseSource
- [ai-chat] added showInChat property to control agent visibility in chat #16925
- [ai-chat] do not open editor on apply change #16986
- [ai-chat] fixed cannot configure tool confirmation mode if default is configured in package.json preferences #17068 - Contributed on behalf of Lonti.com Pty Ltd.
- [ai-chat] fixed: clear image context from chat input after sending #16863
- [ai-chat] improved usability of image file picker #17018
- [ai-chat] small ui improvements #16952
- [ai-chat-ui] added back/forward navigation between chats in the AI Chat View #16894
- [ai-chat-ui] added generic capabilities selection UI for chat requests #17041
- [ai-chat-ui] displayed images in chat request tree nodes #17024
- [ai-chat-ui] improved image variable semantics and behavior #16902 - contributed on behalf of STMicroelectronics
- [ai-chat-ui] improved tool call hover behavior and argument rendering #16990
- [ai-chat-ui] improved tool call parameters rendering #16844
- [ai-chat-ui] used bindRootContributionProvider for GenericCapabilitiesContribution #17070
- [ai-claude-code] implemented session-based tool approval for claude code #16961
- [ai-claude-code] supported for the AskUserQuestion functionality #16981
- [ai-copilot] ux improvements for GitHub copilot integration #17059
- [ai-core] added ./prompts/skills default to preference description #17008
- [ai-core] added capability variable for conditional prompt fragments #16985
- [ai-core] improved AI agent completion notifications #17029 - contributed on behalf of STMicroelectronics
- [ai-core] improved how to enable thinking mode #17023
- [ai-core] made opus 4.6 default in model aliases #17066
- [ai-core] passed toolCallId across RPC and normalized empty tool arguments #16908
- [ai-core] sorted the list by agent name and removed disabled agents #16997
- [ai-core] supported name and description for prompt fragments via frontmatter #17048
- [ai-core] watched parent directory for skills folder creation #16927
- [ai-history] showed prompt variant ID and edit state in history #16876
- [ai-ide] added cards for recent chat sessions to the Welcome screen #16894
- [ai-ide] added MCP server CRUD dialogs and slash commands to skills widget #16991
- [ai-ide] allowed Theia Coder Agent Mode to launch Apps #17012
- [ai-ide] consolidated coder prompts #17004
- [ai-ide] enhanced and refined capabilities #17049
- [ai-ide] fixed fileDiagnosticProvider to not open an editor #17014
- [ai-ide] fixed small improvements in AI agent config view #16974
- [ai-ide] fixed unread badge clearing all sessions when any chat is opened #17045
- [ai-ide] improved AI Chat welcome screen styling and UX #17039
- [ai-ide] made chrome dev App Tester default #17042
- [ai-ide] made new plan mode default #17005
- [ai-ide] Removed confusing escaping comment from function #17006
- [ai-ide] removed usage statistics from devtools mcp #16943
- [ai-ide] sorted agents alphabetically #16979
- [ai-ide] sorted alphabetically the MCP servers, variables, tools, and prompts lists #16983
- [ai-mcp] added roots support for mcp servers #16911
- [ai-ollama] yielded toolCallResponsePart before calling the tool handler #16976
- [ai-terminal] added shell command permission UI for AI terminal tool confirmation #17054
- [ai-terminal] added shell command whitelist for auto-approval #16931
- [ai-terminal] made shell command parser quote-aware #17022
- [api-samples] improved clarity of sample contributions #16937
- [ci] published next version on weekdays instead of only Mondays #16989
- [core] added a simple Card React co

- [ai-chat] added mode selection for coder and architect agents #16860
- [ai-chat] fix: do not break chat when interrupting tool calls #16806
- [ai-chat] fixed refresh bug after manual task context change #16816
- [ai-chat] fixed respect for agentId in delegateToAgent when prompt contains @mentions #16855
- [ai-chat] fixed the failing test #16897
- [ai-chat] implemented preference for AI chat persisted session limit #16776
- [ai-chat] serialized and restored parsedChatRequest in chat sessions #16736
- [ai-chat] stored persistent chat sessions in the workspace by default #16847
- [ai-chat-ui] added toggle button to collapse changeset list #16801
- [ai-chat-ui] displayed prompt variant customization in chat view responses and agent config #16749
- [ai-chat-ui] enhanced chat accessibility (focus commands and ARIA) #16835
- [ai-chat-ui] highlighted referenced tools in users' chat messages #16812
- [ai-copilot] added GitHub Copilot language model integration #16841
- [ai-core] added agent skills support #16810
- [ai-core] updated default models #16828
- [ai-hugging-face] upgraded sdk to v4.x for new api endpoint #16864
- [ai-hugging-face] used patchLanguageModel to update model status #16885
- [ai-ide] added createSkill agent #16903
- [ai-ide] added Theia coder agent mode next prompt #16799
- [ai-ide] added todo tool function #16859
- [ai-ide] appTester improvements #16898
- [ai-ide] fixed misleading tool description for context_addFiles #16778
- [ai-ide] fixed model selection UI update for model alias #16800
- [ai-ide] fixed remote MCP server state and improved UX #16822
- [ai-ide] fixed selected item after model alias change #16760
- [ai-ide] improved architect planning mode - next prompt #16843
- [ai-ide] improved function descriptions #16796
- [ai-ide] improved task context variable description #16817
- [ai-ide] Reverted "Fixed #16795 (#16800)" #16802
- [ai-ide] switched writeFileReplacement to V2 replacer #16815
- [ai-mcp] blocked mcp server autostart in untrusted workspaces #16891
- [ai-mcp] updated MCP SDK to v1.25.1 and migrated to zod v4 #16780
- [core] added injectable LoggerSanitizer to mask sensitive data in logs #16771 - Contributed on behalf of STMicroelectronics
- [core] fixed memory leaks from 'toDisposeOn' disposable collections #16856
- [core] improved keybinding lookup and prioritization #16763
- [core] postponed default override warning logging until builtin preferences registered #16853
- [core] refactored tree search box into react widget and enhanced ux and styling #16761
- [core] restored enhanced tab preview for all widgets #16895
- [core] theia v1.67.0 released #16738
- [core] translation update for version 1.68.0 #16907
- [debug] fixed deletion of disabled source breakpoints #16794
- [debug] improved debug console with output persistence and text filtering #16882
- [doc] updated publishing guide documentation #16768 - Contributed on behalf of STMicroelectronics
- [examples] fixed clean script for the example applications #16767
- [github] added discussion auto-reply #16746
- [github] added sponsoring hint to discussion templates #16745
- [github] added sponsoring hint to gh ticket templates #16744
- [localization-manager] set deepl auth header #16906
- [messages] allowed copying messages from notifications and fixed copying of error messages in AI chat #16830 - Contributed on behalf of STMicroelectronics
- [messages] handled notification timeout correctly when set to 0 #16849 - Contributed on behalf of STMicroelectronics
- [monaco] updated built-in theme and color definitions #16873
- [output] fixed state restoration in outputWidget #16851
- [plugin] preserved lsp diagnostic.data in markers #16766
- [plugin] updated VS Code built-in extensions to 1.104.0 and Typescript to 5.9.3 #16774 - contributed on behalf of STMicroelectronics
- [preferences] marked llmprovider as experimental #16784
- [preferences] supported preference tags in settings view #16783
- [scm] added action button (commit button) support to

- [ai-chat] added mode selection for coder and architect agents #16860
- [ai-chat] fix: do not break chat when interrupting tool calls #16806
- [ai-chat] fixed refresh bug after manual task context change #16816
- [ai-chat] fixed respect for agentId in delegateToAgent when prompt contains @mentions #16855
- [ai-chat] fixed the failing test #16897
- [ai-chat] implemented preference for AI chat persisted session limit #16776
- [ai-chat] serialized and restored parsedChatRequest in chat sessions #16736
- [ai-chat] stored persistent chat sessions in the workspace by default #16847
- [ai-chat-ui] added toggle button to collapse changeset list #16801
- [ai-chat-ui] displayed prompt variant customization in chat view responses and agent config #16749
- [ai-chat-ui] enhanced chat accessibility (focus commands and ARIA) #16835
- [ai-chat-ui] highlighted referenced tools in users' chat messages #16812
- [ai-copilot] added GitHub Copilot language model integration #16841
- [ai-core] added agent skills support #16810
- [ai-core] updated default models #16828
- [ai-hugging-face] upgraded sdk to v4.x for new api endpoint #16864
- [ai-hugging-face] used patchLanguageModel to update model status #16885
- [ai-ide] added createSkill agent #16903
- [ai-ide] added Theia coder agent mode next prompt #16799
- [ai-ide] added todo tool function #16859
- [ai-ide] appTester improvements #16898
- [ai-ide] fixed misleading tool description for context_addFiles #16778
- [ai-ide] fixed model selection UI update for model alias #16800
- [ai-ide] fixed remote MCP server state and improved UX #16822
- [ai-ide] fixed selected item after model alias change #16760
- [ai-ide] improved architect planning mode - next prompt #16843
- [ai-ide] improved function descriptions #16796
- [ai-ide] improved task context variable description #16817
- [ai-ide] Reverted "Fixed #16795 (#16800)" #16802
- [ai-ide] switched writeFileReplacement to V2 replacer #16815
- [ai-mcp] blocked mcp server autostart in untrusted workspaces #16891
- [ai-mcp] updated MCP SDK to v1.25.1 and migrated to zod v4 #16780
- [core] added injectable LoggerSanitizer to mask sensitive data in logs #16771 - Contributed on behalf of STMicroelectronics
- [core] fixed memory leaks from 'toDisposeOn' disposable collections #16856
- [core] improved keybinding lookup and prioritization #16763
- [core] postponed default override warning logging until builtin preferences registered #16853
- [core] refactored tree search box into react widget and enhanced ux and styling #16761
- [core] restored enhanced tab preview for all widgets #16895
- [core] theia v1.67.0 released #16738
- [core] translation update for version 1.68.0 #16907
- [debug] fixed deletion of disabled source breakpoints #16794
- [debug] improved debug console with output persistence and text filtering #16882
- [doc] updated publishing guide documentation #16768 - Contributed on behalf of STMicroelectronics
- [examples] fixed clean script for the example applications #16767
- [github] added discussion auto-reply #16746
- [github] added sponsoring hint to discussion templates #16745
- [github] added sponsoring hint to gh ticket templates #16744
- [localization-manager] set deepl auth header #16906
- [messages] allowed copying messages from notifications and fixed copying of error messages in AI chat #16830 - Contributed on behalf of STMicroelectronics
- [messages] handled notification timeout correctly when set to 0 #16849 - Contributed on behalf of STMicroelectronics
- [monaco] updated built-in theme and color definitions #16873
- [output] fixed state restoration in outputWidget #16851
- [plugin] preserved lsp diagnostic.data in markers #16766
- [plugin] updated VS Code built-in extensions to 1.104.0 and Typescript to 5.9.3 #16774 - contributed on behalf of STMicroelectronics
- [preferences] marked llmprovider as experimental #16784
- [preferences] supported preference tags in settings view #16783
- [scm] added action button (commit button) support to

- [ai-anthropic] added opus 4.5 to default models #16656
- [ai-anthropic, ai-core, ai-google, ai-openai] updated default models #16636
- [ai-chat] clarified role of Orchestrator agent in Theia AI chat interactions #16663
- [ai-chat-ui] ensured active session is recreated after deletion #16702 - Contributed on behalf of Lonti.com Pty Ltd
- [ai-claude-code] added session forking support #16667
- [ai-claude-code] migrated Claude Code to Theia native slash commands and modes #16541
- [ai-codex] added codex as an agent #16484
- [ai-codex] correctly asserted path for windows and linux tests #16620
- [ai-core] enableAgent/disableAgent: added async/await #16599
- [ai-core] enabled markdown syntax for prompt template files #16557
- [ai-core] fixed duplicate command check #16718
- [ai-core] handled undefined parameters for prompt template discard command #16706
- [ai-core] made opus 4.5 default in model alliasses #16734
- [ai-core] returned command by name if not found by ID #16614
- [ai-google] updated @google/genai dependency, added thoughtSignature support and fixed content extraction #16664
- [ai-ide] added GitHub slash commands #16704
- [ai-ide] added remember command #16639
- [ai-ide] clicking on a file in ai context does not open file #16468
- [ai-ide] improved AI Agent Configuration view #16698
- [ai-ide] made next replacer function default #16597
- [ai-ide] updated command prompt template examples #16608
- [ai-llamafile] updated llamafile model status for agent availability #16692
- [ai-openai] recursively strictified tool call schemata #16553
- [ai-terminal] fixed unresponsiveness in ask AI terminal assistant #16714
- [application-package] bumped API compatibility to 1.106.1 #16655 - Contributed on behalf of STMicroelectronics
- [ci] fixed check-new-packages workflow and checkPublish script #16650
- [ci] migrated to npm trusted publishing (OIDC) and updated publishing workflow #16630
- [ci] optimized test builds to avoid redundant executions #16552
- [ci] set default values for scheduled publish-ci runs #16695
- [core] added support for emojis in markdown renderers #16548
- [core] fixed electron startup crash on wayland #16658
- [core] fixed the issue with exclude pattern in user settings not merging with workspace settings #16483
- [core] fixed workbench.startupEditor and scm.defaultViewMode reset behavior #16646
- [core] improved error handling in catalog.json download script #16579 - contributed on behalf of STMicroelectronics
- [core] improved menubar items active state handling on hover #16586 - contributed on behalf of STMicroelectronics
- [core] improved typing related to globalThis object #16603
- [core] npm upgrade #16516
- [core] provided markdown and localizedMarkdown components #16470
- [core] removed references to gitpod (now ona) #16610
- [core] restored tree expansion state preservation in SourceTreeWidget #16654
- [core] reverted "fix(core): restore tree expansion state preservation in SourceTreeWidget" #16672
- [core] reverted to scrollIntoView #16532
- [core] theia v1.66.0 released #16515
- [core] updated nls.metadata for vscode API 1.106.1 and added update eslint localization check #16728 - contributed on behalf of STMicroelectronics
- [core] updated package READMEs #16631
- [core] updated translations for 'applyAll' and 'finished' in Chinese locale #16556
- [core] used PreferenceService.get to merge agent settings from different scopes #16612
- [core] used undefined-safe deep equality check in preference updates #16709
- [customAgents] fixed default model for theia-dev #16680
- [debug] added breakpoint actions to debug view #16700
- [debug] added support for "lazy" debug variables #16681
- [debug] added support for adding the current editor selection to watch #16567
- [debug] added support for data breakpoints #16505
- [debug] opened stack frame editor on tap #16519
- [debug] variables made expandable when they should not #16684
- [dev-container] basic devcontainer docker compose support #16577
- [dev-conta

- [ai-anthropic] allowed configuring proxy settings #16453
- [ai-anthropic] fixed Anthropic request errors when using parallel tool calls #16359
- [ai-chat] enhanced localization of the ai-chat package #16409
- [ai-chat] implemented chat session persistence #16486
- [ai-chat] refined the description of agent delegate tool #16378
- [ai-chat-ui] added mode support for chat agents #16489
- [ai-chat-ui] enhanced localization of the ai-chat-ui package #16414
- [ai-claude-code] added dedicated Claude Code API key preference #16508
- [ai-claude-code] enhanced localization of the ai-claude-code package #16451
- [ai-claude-code] migrated to claude agent sdk #16500
- [ai-code-completion] enhanced localization of ai-code-completion #16416
- [ai-core] added support for slash commands #16444
- [ai-core] enhanced localization of the ai-core package #16350
- [ai-editor] enhanced localization of ai-editor #16416
- [ai-google] conditionally included tools in GoogleModel based on functionDeclarations length #16480
- [ai-google] fixed google language model error for requests without tool functions #16380 - Contributed on behalf of Lonti.com Pty Ltd
- [ai-ide] adapted workspace functions preferences #16452
- [ai-ide] added GitHub agent #16374
- [ai-ide] added initial project info agent (alpha) #16462
- [ai-ide] added new content replacer strategy for coder edit mode #16322
- [ai-ide] added orchestrator agent exclusion list #16510
- [ai-ide] added support to suggest terminal command #16428
- [ai-ide] enhanced localization of the ai-ide package #16426
- [ai-ide] extracted task context prompt into separate agent #16393
- [ai-ide] fixed two minor spelling/wording issues with the coder prompt #16402
- [ai-ide] forbade meta comments in coder prompt #16331
- [ai-ide] removed @theia/git dependency #16465
- [ai-mcp] added option to run a resolve operation on mcp server start #16049
- [ai-mcp] fixed preference categorization for MCP server preferences #16346
- [ai-mcp-server] added .js extension to imports for ESM to allow usage in commonjs #16410 - Contributed by STMicroelectronics
- [ai-openai] allowed configuring proxy settings #16453
- [ai-openai] used OpenAI response API #16394
- [ai-scanoss] fixed SCANOSS dialog expansion with long JSON content #16485
- [ai-terminal] enhanced localization of ai-terminal #16416
- [application-manager] resolved .node files in webpack backend config #16377
- [application-package] bumped vscode API compatibility to 1.105.0 #16495
- [ci] optimized CI build to avoid redundant executions #16457
- [ci] updated license workflow and switched to nodejs wrapper #16456
- [ci] updated publish-release workflow for release automation #16433 - Contributed on behalf of STMicroelectronics
- [core] added symbol icon default colors #15860
- [core] added z-index to dock layering style #16375
- [core] ensured reveal scrolls to selected tree row #16463
- [core] evaluated enablement and toggle state of command only once before showing context menu #16325 - Contributed on behalf of Lonti.com Pty Ltd
- [core] evolved vscode API (public) to 1.105.0 #16476 - Contributed on behalf of STMicroelectronics
- [core] fixed authentication service not reacting to session changes #16252
- [core] fixed tooltip/hover service mouseOut for non-chromium browsers #16417
- [core] fixed vscode plugin activation failed error when it uses an already existing key in its settings #16481
- [core] improved default value and reset handling for preferences #16356
- [core] prevented await identifiers #16404 - Contributed on behalf of STMicroelectronics
- [core] showed hover tooltip immediately when clicking status bar item #16385
- [core] updated toggle developer tools keyboard shortcut to avoid overlap #16488
- [core] upgraded NPM #16357
- [debug] added check for valid debug session #16330
- [debug] added support for context menu in debug hover #16366
- [debug] fixed Add to Watch command #16340
- [debug] made reveal position smoother #16311 - Contributed on behalf of MVT

- [ai-anthropic] implemented prompt caching for Anthropic #15731
- [ai-anthropic] removed tool results from subsequential requests #15702
- [ai-anthropic] reverted "fix: remove tool results from subsequential requests" #15721
- [ai-anthropic] set opus 4 to 32000 max tokens #15788
- [ai-chat] adapted all prompt ids to common schema #15880
- [ai-chat] added async initialization to changesetfileelement #15761
- [ai-chat] enabled agent-to-agent delegation via tool calls #15736
- [ai-chat] filtered out error messages from the messages sent to the llm #15720
- [ai-chat] fixed change set deletion #15759
- [ai-chat] fixed pinning of the last mentioned agent rather than the first one #15777
- [ai-chat] introduced tool call confirmation UI #15714
- [ai-chat] removed requirement that agent has to be mentioned first #15854
- [ai-chat] sanitized task context labels #15762
- [ai-chat] provided a task context management / agent #15732
- [ai-chat-ui] added chat retry button #15779
- [ai-chat-ui] fixed chat input widget not tracking branches and pending state #15727
- [ai-chat-ui] fixed difficulty to scroll up when auto scroll is enabled #15748
- [ai-chat-ui] fixed paste handling in chat input #15766
- [ai-chat-ui] fixed temporary scroll lock when user scrolls up in AI Chat #15683
- [ai-chat-ui] fixed toolcall spinner #15757
- [ai-chat-ui] showed active chat in chat dropdown #15723
- [ai-chat-ui] started chat from task context #15700
- [ai-chat-ui] supported instant request for chat initialized with task context and agent #15778
- [ai-chat-ui] replaced YOLO terminology with "Always Allow" #15756
- [ai-code-completion] added support for custom AI variables in code completion prompt #15681 - Contributed on behalf of Lonti.com Pty Ltd.
- [ai-code-completion] fixed caching in code completion #15855
- [ai-code-completion] worked on improving code completion #15715
- [ai-core] added chat variable for all open editors #15775
- [ai-core] added fall back to default prompts if selected does not exist #15879
- [ai-core] AI features accessible when AI is not enabled #15780
- [ai-core] allowed to increase the number of retries before failing #15730
- [ai-core] allowed adding variants via files #15815
- [ai-core] fixed bug that customizations are treated as variants #15846
- [ai-core] handled toolCall errors gracefully instead of throwing #15800
- [ai-core] implemented theia-ai agent completion notification #15816
- [ai-core] improved image support for tool calls #15765
- [ai-core] fixed prompt fragment view resetting custom prompt fragments #15852
- [ai-editor] improved ask AI feature #15725
- [ai-editor] refined inline Ask AI chat input appearance #15885
- [ai-editor] started chat from editor context #15712
- [ai-editor] used chatInputWidget in AskAI Input and fixed keyboard scrolling #15781
- [ai-editor] streamlined ask AI commands and fixed send to AI chat error #15839
- [ai-google] updated the Google AI SDK #15737
- [ai-ide] added ai tools for write through file changes #15717
- [ai-ide] consolidated coder prompts #15733
- [ai-ide] fixed agent tree selection highlighting in AI configuration widget #15850
- [ai-ide] fixed AI Configuration widget tab bar overflow #15739
- [ai-ide] included the ai-ide stylesheet in the secondary window #15752
- [ai-ide] introduced appTester agent via browser automation #15713
- [ai-ide] made workspace functions token efficient #15703
- [ai-ide] made workspace search function more efficient / use less tokens #15743
- [ai-ide] moved AI configuration view to main #15726
- [ai-ide] refined coder prompt #15774
- [ai-ide] reminded coder to use file change functions #15848
- [ai-ide] reorganized architect prompts #15859
- [ai-ide] used relative path in search results in workspace function #15704
- [ai-ide] workspace search function allowed specifying file extensions #15699
- [ai-ollama] updated ollama version and supported newest features #15795
- [ai-openai] made sure to add openai stream options only in streaming models #15706

Update to latest Theia