0.2.4 release #549
0.2.4 release #549
Conversation
refactor: migration to pnpm
fix: fix sharp on mac arm build by x64
added togetherAI provider : TogetherAI provider
feat: tray menu check version
add oxlint for default lint
feat: show laste-version-dialog tips for user
Add TogetherAI Provider
Co-authored-by: duskzhen <zerob13@gmail.com>
feat: setting show/hide key
feat: enhance model selection logic in common setting
…sion feat: add support for mcp http 400
fix(markdown): initial style undefined
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the You can disable this status message by setting the WalkthroughThis update introduces centralized model configuration management with persistent storage, dynamic model type filtering, and batch status retrieval. It adds new LLM providers (Minimax, Together), dynamic model discovery for Gemini, and OpenRouter/PPIO API-driven model metadata. The Powerpack server now supports E2B sandboxed Python execution. UI components and localization are extended for model configuration and E2B integration. The build system migrates to pnpm, and session recovery logic is enhanced for MCP clients. Changes
Sequence Diagram(s)sequenceDiagram
participant User
participant UI (ModelConfigDialog)
participant Main Process (ConfigPresenter)
participant ModelConfigHelper (Store)
User->>UI (ModelConfigDialog): Open dialog for model config
UI (ModelConfigDialog)->>Main Process (ConfigPresenter): getModelConfig(modelId, providerId)
Main Process (ConfigPresenter)->>ModelConfigHelper (Store): getModelConfig(modelId, providerId)
ModelConfigHelper (Store)-->>Main Process (ConfigPresenter): Return config (user/provider/default)
Main Process (ConfigPresenter)-->>UI (ModelConfigDialog): Return config
User->>UI (ModelConfigDialog): Edit and save config
UI (ModelConfigDialog)->>Main Process (ConfigPresenter): setModelConfig(modelId, providerId, config)
Main Process (ConfigPresenter)->>ModelConfigHelper (Store): setModelConfig(modelId, providerId, config)
ModelConfigHelper (Store)-->>Main Process (ConfigPresenter): Ack
Main Process (ConfigPresenter)-->>UI (ModelConfigDialog): Ack (emit config changed event)
sequenceDiagram
participant MCP Client
participant MCP Server
MCP Client->>MCP Server: Tool call / List / Get
MCP Server-->>MCP Client: Error (session invalid)
MCP Client->>MCP Client: Detect session error
MCP Client->>MCP Client: Restart service, clear cache
MCP Client->>MCP Server: Retry operation
MCP Server-->>MCP Client: (If error again) Error (session invalid)
MCP Client->>MCP Client: Stop service, emit stopped event
sequenceDiagram
participant User
participant UI (mcpServerForm)
participant Main Process (PowerpackServer)
User->>UI (mcpServerForm): Enable E2B, enter API key
UI (mcpServerForm)->>Main Process (PowerpackServer): Start server with USE_E2B and E2B_API_KEY
Main Process (PowerpackServer)->>Main Process (PowerpackServer): Setup E2B config
Main Process (PowerpackServer)->>Main Process (PowerpackServer): Expose run_code tool for E2B
User->>UI (mcpServerForm): Run Python code
UI (mcpServerForm)->>Main Process (PowerpackServer): Execute run_code
Main Process (PowerpackServer)->>E2B Sandbox: Run code
E2B Sandbox-->>Main Process (PowerpackServer): Return result
Main Process (PowerpackServer)-->>UI (mcpServerForm): Show execution result
Possibly related PRs
Poem
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
CodeRabbit Configuration File (
|
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
There was a problem hiding this comment.
Actionable comments posted: 7
🧹 Nitpick comments (21)
src/main/presenter/llamaCppPresenter/llama.ts (1)
147-147: Remove or replace the stray console.log.This line appears to be a leftover debug statement that will log
'llama'on every load. Please remove it or replace it with a structured logger at the appropriate level.pnpm-workspace.yaml (1)
1-5: Consider the implications of using "current" architecture only.The configuration restricts support to the current OS and CPU architecture only. While this works for single-platform development, it may cause issues in:
- Cross-platform development scenarios
- CI/CD pipelines running on different architectures
- Team members using different system architectures
Consider if this restriction aligns with your project's deployment and development needs.
If broader architecture support is needed, you could specify explicit architectures:
supportedArchitectures: os: - - current + - win32 + - darwin + - linux cpu: - - current + - x64 + - arm64electron-builder.yml (1)
8-8: Remove duplicate test directory exclusion.The test directory exclusion is correctly added, but there's a duplicate entry on line 19. Having the same exclusion pattern twice is unnecessary.
Apply this diff to remove the duplicate:
- '!electron-builder.yml' - '!electron-builder-macx64.yml' - - '!test/*' - '!*.config.ts'src/renderer/src/components/editor/mention/MentionList.vue (2)
19-19: Inconsistent<Icon>formatting.
You’ve inlined this<Icon>component, whereas others use multi-line props. Consider standardizing either single-line for short tags or multi-line when there are multiple props.
74-80: SimplifyhasFilescheck.
You can shrink the boolean logic to a one-liner using optional chaining andArray.isArray:return Array.isArray(item.mcpEntry?.files) && item.mcpEntry.files.length > 0electron-builder-macx64.yml (1)
7-9: Duplicate and too narrow exclusion pattern for tests.
The pattern!test/*appears twice and only excludes top-level test files. Consolidate into one entry and use a recursive glob, e.g.:- - '!test/*' - - '!test/*' + - '!**/test/**'Also applies to: 18-20
README.md (1)
272-273: Fix grammatical issue in Windows note.The note is helpful for Windows users, but there's a minor grammatical issue identified by static analysis.
Apply this diff to fix the grammar:
-* For Windows: To allow non-admin users to create symlinks and hardlinks, enable `Developer Mode` in Settings or use an administrator account. Otherwise `pnpm` ops will fail. +* For Windows: To allow non-admin users to create symlinks and hardlinks, enable `Developer Mode` in Settings or use an administrator account. Otherwise, `pnpm` ops will fail.README.jp.md (1)
268-268: Fix markdown formatting issues.Static analysis identified minor markdown formatting inconsistencies that should be addressed for better compliance.
Apply these fixes:
-* For Windows: 非管理者ユーザーがシンボリックリンクやハードリンクを作成できるようにするには、設定で「開発者モード」を有効にするか、管理者アカウントを使用してください。それ以外の場合、pnpm の操作は失敗します。 +- For Windows: 非管理者ユーザーがシンボリックリンクやハードリンクを作成できるようにするには、設定で「開発者モード」を有効にするか、管理者アカウントを使用してください。それ以外の場合、pnpm の操作は失敗します。And for consistency with markdown best practices, consider adding output examples for shell commands where appropriate.
Also applies to: 273-273
src/renderer/src/components/NewThread.vue (1)
236-248: Improve model selection logic efficiencyThe current implementation flattens all models from all providers and then searches for the first matching type. This approach works but could be optimized.
Consider this more efficient approach that avoids unnecessary array operations:
- const model = settingsStore.enabledModels - .flatMap((provider) => - provider.models.map((m) => ({ ...m, providerId: provider.providerId })) - ) - .find((m) => m.type === ModelType.Chat || m.type === ModelType.ImageGeneration) + let model = null + outerLoop: for (const provider of settingsStore.enabledModels) { + for (const m of provider.models) { + if (m.type === ModelType.Chat || m.type === ModelType.ImageGeneration) { + model = { ...m, providerId: provider.providerId } + break outerLoop + } + } + }This approach stops searching once the first matching model is found, avoiding unnecessary iterations.
scripts/install-sharp-for-platform.js (1)
28-35: Minor formatting and consistency issue in Linux configuration.There's a missing space after the comma in the Linux ARM64 configuration and an inconsistent comment about WebAssembly support.
Apply this diff to fix the formatting:
'linux-arm64': { - os: ['current','linux'], + os: ['current', 'linux'], cpu: ['current', 'wasm32'], },Also, consider clarifying why
wasm32is included for Linux platforms but not others, or make the comment more specific about Sharp's WebAssembly requirements for Linux.src/renderer/src/components/settings/ModelConfigDialog.vue (1)
254-284: Validation logic is sound but consider consolidating error checking.The validation function checks all constraints properly, but the computed
isValidproperty callsvalidateForm()every time it's accessed, which could be inefficient.Consider using a reactive approach to validation:
-// 表单是否有效 -const isValid = computed(() => { - validateForm() - return Object.keys(errors.value).length === 0 -}) +// 表单是否有效 +const isValid = computed(() => { + return Object.keys(errors.value).length === 0 +}) + +// 在 config 变化时自动验证 +watch(config, () => { + validateForm() +}, { deep: true })src/main/presenter/llmProviderPresenter/providers/minimaxProvider.ts (2)
6-9: Remove the unnecessary constructor.The constructor doesn't add any functionality beyond calling the parent constructor.
Apply this diff to remove the unnecessary constructor:
- constructor(provider: LLM_PROVIDER, configPresenter: ConfigPresenter) { - // 初始化智谱AI模型配置 - super(provider, configPresenter) - }
47-64: Consider localization consistency for summary prompts.The summary prompt is hardcoded in English, which may not align with the application's localization strategy. Consider making this configurable or using the user's language preference.
src/main/presenter/llmProviderPresenter/providers/togetherProvider.ts (1)
6-8: Remove the unnecessary constructor.The constructor doesn't add any functionality beyond calling the parent constructor.
Apply this diff to remove the unnecessary constructor:
- constructor(provider: LLM_PROVIDER, configPresenter: ConfigPresenter) { - super(provider, configPresenter) - }src/main/presenter/llmProviderPresenter/providers/ppioProvider.ts (1)
144-229: Consider removing or conditionalizing the console.log statement.The implementation looks solid with proper feature extraction and configuration updates. However, line 222 contains an active console.log that might be too verbose for production.
Consider wrapping the log in a debug condition:
-console.log(`Processed ${models.length} PPIO models with dynamic configuration updates`) +if (process.env.DEBUG_PPIO) { + console.log(`Processed ${models.length} PPIO models with dynamic configuration updates`) +}docs/mcp-session-recovery.md (1)
1-241: Add language specifiers to fenced code blocks.The documentation is comprehensive and well-written. However, some code blocks are missing language specifiers which would improve syntax highlighting.
Add language specifiers to the code blocks:
- Line 7: Add language specifier
- Line 175: Add
textorlogspecifier- Line 181: Add
textorlogspecifierExample:
-``` +```text Error POSTing to endpoint (HTTP 400): Bad Request: No valid session ID providedsrc/main/presenter/llmProviderPresenter/index.ts (1)
647-863: Well-structured refactoring of tool call handling.The distinction between native and non-native function calling scenarios is clearly implemented. The approach of formatting tool records as XML for non-native providers aligns well with the prompt changes in baseProvider.ts.
Consider internationalizing the Chinese prompt texts for broader applicability.
Consider making the prompt texts configurable or translatable:
-const userPromptText = '以上是你刚执行的工具调用及其响应信息，已帮你插入，请仔细阅读工具响应，并继续你的回答。' +const locale = this.configPresenter.getLanguage?.() || 'zh-CN' +const userPromptText = this.getToolResponsePrompt(locale)src/main/presenter/llmProviderPresenter/providers/openRouterProvider.ts (1)
191-191: Remove commented debug statementsDebug logging statements should be removed from production code to reduce noise and improve maintainability.
- // console.log('OpenRouter models response:', JSON.stringify(response, null, 2)) - // Update configuration if changed if (configChanged) { - // console.log(`Updating OpenRouter configuration for model ${modelId}:`, { - // old: { - // contextLength: existingConfig.contextLength, - // maxTokens: existingConfig.maxTokens, - // functionCall: existingConfig.functionCall, - // vision: existingConfig.vision, - // reasoning: existingConfig.reasoning - // }, - // new: newConfig - // }) - this.configPresenter.setModelConfig(modelId, this.provider.id, newConfig) }Also applies to: 245-254
src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts (1)
335-349: Add logging for auto-enabled modelsConsider adding more detailed logging about which specific matching strategy was used for each auto-enabled model to help with debugging and monitoring.
const modelsToEnable = this.models.filter((model) => { - return this.isModelRecommended(model.id, recommendedModelIds) + const isRecommended = this.isModelRecommended(model.id, recommendedModelIds) + if (isRecommended) { + console.debug(`Model ${model.id} matched as recommended model`) + } + return isRecommended })src/main/presenter/mcpPresenter/mcpClient.ts (2)
281-281: Use debug level for verbose logging.Consider using
console.debuginstead ofconsole.infoandconsole.logfor verbose logging that's primarily useful during development.Apply this diff:
- console.info('runtimeBasePath', runtimeBasePath) + console.debug('runtimeBasePath', runtimeBasePath)- console.log('mcp env', command) + console.debug('mcp env', command)Also applies to: 518-518
667-716: Robust session recovery implementation.The session recovery mechanism is well-designed with proper safeguards against infinite loops. The single retry approach is simple and effective.
Consider future enhancements:
- Exponential backoff for multiple retry attempts
- Configurable retry limits
- Metrics/telemetry for recovery attempts
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (107)
.cursor/rules/build-deploy.mdc(0 hunks).cursor/rules/development-setup.mdc(1 hunks).github/workflows/build.yml(4 hunks).github/workflows/prcheck.yml(1 hunks).oxlintrc.json(1 hunks).prettierignore(1 hunks)CONTRIBUTING.md(3 hunks)CONTRIBUTING.zh.md(3 hunks)README.jp.md(1 hunks)README.md(1 hunks)README.zh.md(1 hunks)docs/config-presenter-design.md(0 hunks)docs/developer-guide.md(5 hunks)docs/mcp-session-recovery.md(1 hunks)docs/user-guide.md(1 hunks)electron-builder-macx64.yml(1 hunks)electron-builder.yml(1 hunks)eslint.config.mjs(0 hunks)package.json(4 hunks)pnpm-workspace.yaml(1 hunks)scripts/install-sharp-for-platform.js(1 hunks)src/main/eventbus.ts(1 hunks)src/main/events.ts(2 hunks)src/main/index.ts(1 hunks)src/main/presenter/configPresenter/index.ts(8 hunks)src/main/presenter/configPresenter/mcpConfHelper.ts(1 hunks)src/main/presenter/configPresenter/modelConfig.ts(1 hunks)src/main/presenter/configPresenter/modelDefaultSettings.ts(3 hunks)src/main/presenter/configPresenter/providerModelSettings.ts(2 hunks)src/main/presenter/configPresenter/providers.ts(1 hunks)src/main/presenter/filePresenter/CsvFileAdapter.ts(1 hunks)src/main/presenter/filePresenter/FilePresenter.ts(1 hunks)src/main/presenter/githubCopilotDeviceFlow.ts(4 hunks)src/main/presenter/llamaCppPresenter/llama.ts(1 hunks)src/main/presenter/llmProviderPresenter/baseProvider.ts(2 hunks)src/main/presenter/llmProviderPresenter/index.ts(9 hunks)src/main/presenter/llmProviderPresenter/oauthHelper.ts(1 hunks)src/main/presenter/llmProviderPresenter/providers/anthropicProvider.ts(3 hunks)src/main/presenter/llmProviderPresenter/providers/deepseekProvider.ts(2 hunks)src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts(8 hunks)src/main/presenter/llmProviderPresenter/providers/githubCopilotProvider.ts(0 hunks)src/main/presenter/llmProviderPresenter/providers/minimaxProvider.ts(1 hunks)src/main/presenter/llmProviderPresenter/providers/openAICompatibleProvider.ts(2 hunks)src/main/presenter/llmProviderPresenter/providers/openAIResponsesProvider.ts(1 hunks)src/main/presenter/llmProviderPresenter/providers/openRouterProvider.ts(7 hunks)src/main/presenter/llmProviderPresenter/providers/ppioProvider.ts(4 hunks)src/main/presenter/llmProviderPresenter/providers/siliconcloudProvider.ts(1 hunks)src/main/presenter/llmProviderPresenter/providers/togetherProvider.ts(1 hunks)src/main/presenter/mcpPresenter/inMemoryServers/builder.ts(1 hunks)src/main/presenter/mcpPresenter/inMemoryServers/customPromptsServer.ts(1 hunks)src/main/presenter/mcpPresenter/inMemoryServers/filesystem.ts(1 hunks)src/main/presenter/mcpPresenter/inMemoryServers/powerpackServer.ts(11 hunks)src/main/presenter/mcpPresenter/index.ts(2 hunks)src/main/presenter/mcpPresenter/mcpClient.ts(25 hunks)src/main/presenter/mcpPresenter/serverManager.ts(4 hunks)src/main/presenter/notifactionPresenter.ts(1 hunks)src/main/presenter/proxyConfig.ts(1 hunks)src/main/presenter/shortcutPresenter.ts(0 hunks)src/main/presenter/sqlitePresenter/importData.ts(1 hunks)src/main/presenter/syncPresenter/index.ts(13 hunks)src/main/presenter/tabPresenter.ts(1 hunks)src/main/presenter/threadPresenter/contentEnricher.ts(2 hunks)src/main/presenter/threadPresenter/index.ts(1 hunks)src/main/presenter/threadPresenter/messageManager.ts(2 hunks)src/main/presenter/trayPresenter.ts(1 hunks)src/main/presenter/upgradePresenter/index.ts(4 hunks)src/main/presenter/windowPresenter/index.ts(1 hunks)src/renderer/src/assets/style.css(2 hunks)src/renderer/src/components/ModelSelect.vue(3 hunks)src/renderer/src/components/NewThread.vue(3 hunks)src/renderer/src/components/SideBar.vue(2 hunks)src/renderer/src/components/TitleView.vue(2 hunks)src/renderer/src/components/editor/mention/MentionList.vue(4 hunks)src/renderer/src/components/editor/mention/PromptParamsDialog.vue(1 hunks)src/renderer/src/components/editor/mention/suggestion.ts(1 hunks)src/renderer/src/components/mcp-config/mcpServerForm.vue(10 hunks)src/renderer/src/components/settings/AboutUsSettings.vue(1 hunks)src/renderer/src/components/settings/CommonSettings.vue(3 hunks)src/renderer/src/components/settings/ModelConfigDialog.vue(1 hunks)src/renderer/src/components/settings/ModelConfigItem.vue(3 hunks)src/renderer/src/components/settings/ModelProviderSettingsDetail.vue(2 hunks)src/renderer/src/components/settings/PromptSetting.vue(7 hunks)src/renderer/src/components/settings/ProviderApiConfig.vue(3 hunks)src/renderer/src/components/settings/ProviderModelList.vue(4 hunks)src/renderer/src/components/settings/ProviderModelManager.vue(2 hunks)src/renderer/src/components/settings/ShortcutSettings.vue(2 hunks)src/renderer/src/components/ui/UpdateDialog.vue(1 hunks)src/renderer/src/i18n/en-US/contextMenu.json(1 hunks)src/renderer/src/i18n/en-US/settings.json(3 hunks)src/renderer/src/i18n/en-US/update.json(1 hunks)src/renderer/src/i18n/fa-IR/settings.json(3 hunks)src/renderer/src/i18n/fa-IR/update.json(1 hunks)src/renderer/src/i18n/fr-FR/contextMenu.json(1 hunks)src/renderer/src/i18n/fr-FR/index.ts(1 hunks)src/renderer/src/i18n/fr-FR/settings.json(3 hunks)src/renderer/src/i18n/fr-FR/update.json(1 hunks)src/renderer/src/i18n/ja-JP/contextMenu.json(1 hunks)src/renderer/src/i18n/ja-JP/settings.json(3 hunks)src/renderer/src/i18n/ja-JP/update.json(1 hunks)src/renderer/src/i18n/ko-KR/contextMenu.json(1 hunks)src/renderer/src/i18n/ko-KR/settings.json(3 hunks)src/renderer/src/i18n/ko-KR/update.json(1 hunks)src/renderer/src/i18n/ru-RU/settings.json(3 hunks)src/renderer/src/i18n/ru-RU/update.json(1 hunks)src/renderer/src/i18n/zh-CN/contextMenu.json(1 hunks)src/renderer/src/i18n/zh-CN/settings.json(3 hunks)src/renderer/src/i18n/zh-CN/update.json(1 hunks)
⛔ Files not processed due to max files limit (22)
- src/renderer/src/i18n/zh-HK/settings.json
- src/renderer/src/i18n/zh-HK/update.json
- src/renderer/src/i18n/zh-TW/contextMenu.json
- src/renderer/src/i18n/zh-TW/settings.json
- src/renderer/src/i18n/zh-TW/update.json
- src/renderer/src/stores/mcp.ts
- src/renderer/src/stores/settings.ts
- src/renderer/src/stores/upgrade.ts
- src/renderer/src/views/WelcomeView.vue
- src/shared/i18n.ts
- src/shared/model.ts
- src/shared/presenter.d.ts
- test/README.md
- test/main/eventbus/eventbus.test.ts
- test/main/presenter/llmProviderPresenter.test.ts
- test/main/presenter/mcpClient.test.ts
- test/main/presenter/modelConfig.test.ts
- test/renderer/shell/main.test.ts
- test/setup.renderer.ts
- test/setup.ts
- vitest.config.renderer.ts
- vitest.config.ts
💤 Files with no reviewable changes (5)
- .cursor/rules/build-deploy.mdc
- src/main/presenter/llmProviderPresenter/providers/githubCopilotProvider.ts
- src/main/presenter/shortcutPresenter.ts
- eslint.config.mjs
- docs/config-presenter-design.md
🧰 Additional context used
🪛 markdownlint-cli2 (0.17.2)
docs/developer-guide.md
13-13: Link fragments should be valid
null
(MD051, link-fragments)
README.jp.md
268-268: Unordered list style
Expected: dash; Actual: asterisk
(MD004, ul-style)
273-273: Dollar signs used before commands without showing output
null
(MD014, commands-show-output)
docs/mcp-session-recovery.md
7-7: Fenced code blocks should have a language specified
null
(MD040, fenced-code-language)
175-175: Fenced code blocks should have a language specified
null
(MD040, fenced-code-language)
181-181: Fenced code blocks should have a language specified
null
(MD040, fenced-code-language)
🪛 LanguageTool
README.md
[uncategorized] ~272-~272: A comma may be missing after the conjunctive/linking adverb ‘Otherwise’.
Context: ...ttings or use an administrator account. Otherwise pnpm ops will fail. ### Start Develo...
(SENT_START_CONJUNCTIVE_LINKING_ADVERB_COMMA)
docs/mcp-session-recovery.md
[uncategorized] ~158-~158: 动词的修饰一般为‘形容词（副词）+地+动词’。您的意思是否是：新"地"连接
Context: ...用者知道需要重试 5. 重新调用: 上层调用者重新发起请求，此时会建立新的连接和session 6. 持续错误检测: 如果重启后再次出现sessi...
(wb4)
🪛 YAMLlint (1.37.1)
.github/workflows/prcheck.yml
[error] 49-49: no new line character at the end of file
(new-line-at-end-of-file)
🪛 Biome (1.9.4)
src/main/presenter/llmProviderPresenter/providers/minimaxProvider.ts
[error] 6-11: This constructor is unnecessary.
Unsafe fix: Remove the unnecessary constructor.
(lint/complexity/noUselessConstructor)
src/main/presenter/llmProviderPresenter/providers/togetherProvider.ts
[error] 6-8: This constructor is unnecessary.
Unsafe fix: Remove the unnecessary constructor.
(lint/complexity/noUselessConstructor)
🔇 Additional comments (195)
src/main/presenter/llmProviderPresenter/providers/openAIResponsesProvider.ts (1)
568-573: Formatting consistency forusagetype
The multi-line type annotation forusageis now properly indented, improving readability without affecting behavior.src/renderer/src/i18n/ko-KR/contextMenu.json (1)
17-17: Remove trailing space and add newline at EOF.This change is purely formatting—no content modifications. Ensures consistency across localization files by ending with a newline.
src/renderer/src/assets/style.css (2)
80-82: Approve whitespace cleanup in default:root.Removed trailing spaces after CSS variable definitions for
--usage-low,--usage-mid, and--usage-high. No functional impact, just formatting.
120-122: Approve whitespace cleanup in.darktheme.Removed trailing spaces after CSS variable definitions for
--usage-low,--usage-mid, and--usage-highin the dark theme. Purely formatting change.src/renderer/src/i18n/zh-CN/contextMenu.json (1)
17-17: Skip formatting-only update
Removed trailing whitespace and added a newline. No changes to JSON keys or values.src/renderer/src/i18n/fr-FR/contextMenu.json (1)
17-17: Skip formatting-only update
Removed trailing whitespace and added a newline. No changes to JSON keys or values.src/renderer/src/i18n/en-US/contextMenu.json (1)
17-17: Skip formatting-only update
Removed trailing whitespace and added a newline. No changes to JSON keys or values.src/main/presenter/llmProviderPresenter/oauthHelper.ts (1)
16-16: Skip formatting-only adjustment
Removed an extra space inside the constructor’s braces. No functional changes introduced.docs/user-guide.md (1)
141-141: ```shell
#!/bin/bashSearch the entire repository for any exact occurrences of the old MCP expansion
rg -n "Model Controller Platform"
Also perform a case-insensitive search to catch any variations
rg -i -n "Model Controller Platform"
</details> <details> <summary>src/main/presenter/filePresenter/CsvFileAdapter.ts (1)</summary> `23-23`: **Improved CSV row filtering logic.** The change correctly refines the filtering condition to only exclude rows where all cells are empty, rather than excluding empty arrays. This better handles CSV parsing edge cases where rows might contain empty string cells but still be structurally valid. </details> <details> <summary>src/main/presenter/filePresenter/FilePresenter.ts (1)</summary> `208-211`: **Clean code improvement by removing unused error parameter.** The error parameter was not being used in the catch block, so removing it improves code clarity while maintaining the same functionality. </details> <details> <summary>src/renderer/src/i18n/en-US/update.json (1)</summary> `13-15`: **Well-structured localization additions for update status.** The new localization keys provide clear, user-friendly messages for update scenarios: - `"restarting"` for restart status - `"alreadyUpToDate"` for concise status indication - `"alreadyUpToDateDesc"` for detailed explanation The text is clear and provides good user experience for update status feedback. </details> <details> <summary>src/renderer/src/i18n/ko-KR/update.json (1)</summary> `14-15`: **LGTM! Korean localization strings are accurate and well-translated.** The Korean translations are grammatically correct and contextually appropriate for update status messaging. The consistency with similar additions across other localization files demonstrates good internationalization practices. </details> <details> <summary>src/main/presenter/threadPresenter/contentEnricher.ts (2)</summary> `333-333`: **Good cleanup: Removed unused error parameter.** The simplified catch block improves code clarity while maintaining the same fallback behavior for URL construction failures. --- `351-351`: **Good cleanup: Removed unused error parameter.** Consistent with the previous catch block, this simplification maintains functionality while improving code readability. </details> <details> <summary>docs/developer-guide.md (1)</summary> `13-13`: **LGTM! Correct terminology update for MCP.** The change from "Model Controller Platform" to "Model Context Protocol" is accurate and consistent across all references in the documentation. However, please verify that the table of contents link fragment on line 13 correctly targets the updated heading: ```shell #!/bin/bash # Description: Verify the TOC link fragment matches the updated heading # Expected: The link fragment should match the corrected heading format # Check if the heading exists and matches the link fragment rg -A 2 -B 2 "Model Context Protocol.*MCP" docs/developer-guide.mdAlso applies to: 61-61, 74-74, 76-76, 88-88, 131-131
src/renderer/src/components/editor/mention/suggestion.ts (1)
32-45: Excellent formatting improvement for complex type annotation.The multiline format significantly improves readability of the complex function type without changing its functionality. This aligns well with the overall code quality improvements in this release.
.prettierignore (1)
9-23: Well-structured Prettier ignore additions.The new ignore patterns appropriately exclude:
- Project directories that shouldn't be formatted (docs, build, resources)
- Configuration files with specific formatting requirements (tailwind.config.js, electron.vite.config.ts)
- UI component libraries that should maintain original formatting
- Environment and container files with their own conventions
- IDE/tooling directories
These additions align well with the project's tooling standardization and maintain clear formatting boundaries.
.oxlintrc.json (1)
1-17: Configuration looks solid.
The ignorePatterns cover all the major build and tooling directories. No issues detected.src/renderer/src/i18n/ja-JP/contextMenu.json (1)
1-17: No functional changes.
Only formatting (newline added). Nothing to review here.src/main/presenter/notifactionPresenter.ts (1)
39-43: Formatting alignment approved.
The multi-lineeventBus.sendToRenderernow matches other presenter calls and explicitly usesSendTarget.ALL_WINDOWS.src/renderer/src/components/settings/AboutUsSettings.vue (1)
190-190: LGTM! API signature update maintains consistency.The addition of the boolean parameter to
upgrade.checkUpdate(false)aligns with the broader API update pattern seen across UI components. Thefalseparameter correctly indicates this is a user-initiated update check rather than a silent background check.src/renderer/src/components/SideBar.vue (1)
100-100: LGTM! Clear semantic distinction in update check parameters.The boolean parameters provide logical separation:
checkUpdate(false)for user-initiated checks (line 100)checkUpdate(true)for automatic/silent checks (line 121)This approach maintains consistency with the broader API update and provides clear intent for different update check scenarios.
Also applies to: 121-121
src/main/presenter/llmProviderPresenter/providers/siliconcloudProvider.ts (2)
105-105: LGTM! Header key formatting improvement.Removing quotes from the
Authorizationheader key improves code readability while maintaining functionality. This is consistent with modern JavaScript/TypeScript style practices.
112-114: LGTM! Error message formatting enhancement.The multi-line template literal formatting improves readability of the error message without changing the functional behavior. This makes debugging easier when API key checks fail.
src/main/presenter/mcpPresenter/inMemoryServers/customPromptsServer.ts (1)
88-91: LGTM! Indentation improvement enhances readability.The improved indentation of object properties within the map callback enhances code readability and maintains consistent formatting standards without affecting functionality.
src/main/presenter/trayPresenter.ts (2)
43-43: LGTM! Menu label simplification.Removing the shortcut hint from the tray menu label improves visual clarity while maintaining functionality. The label remains descriptive with
labels.open || '打开/隐藏'.
48-53: LGTM! Well-integrated update check menu item.The new "检查更新" menu item provides convenient access to update functionality directly from the system tray. The implementation correctly:
- Uses localized labels with fallback
- Triggers the appropriate
CHECK_FOR_UPDATESevent via eventBus- Follows the established event-driven architecture pattern
This integrates seamlessly with the broader update check system enhancements.
src/renderer/src/components/editor/mention/PromptParamsDialog.vue (1)
114-116: LGTM! Clean formatting improvements.The added blank line and parentheses around the arrow function parameter improve code readability and consistency without changing the logic.
src/renderer/src/i18n/fr-FR/update.json (1)
14-15: LGTM! Proper French localization additions.The new keys provide clear messaging for when the application is already up-to-date. The French translations are grammatically correct and consistent with the existing localization pattern.
src/renderer/src/i18n/fr-FR/index.ts (1)
31-31: LGTM! Improved string readability.Changing from single quotes with escaping to double quotes makes the disclaimer text much more readable without affecting functionality.
src/main/eventbus.ts (1)
51-51: LGTM! Critical method name fix.Correcting
sendTodefaultTabtosendToDefaultTabfixes the camelCase naming and ensures proper event dispatching to the default tab window.CONTRIBUTING.zh.md (1)
23-23: LGTM! Consistent package manager migration.The documentation has been correctly updated to reflect the migration from npm to pnpm across all command examples. The changes are consistent with the PR objective of officially switching package management to pnpm.
Also applies to: 87-88, 93-93, 113-115
src/renderer/src/components/TitleView.vue (2)
77-77: LGTM! Proper import of ModelType enum.The import statement correctly brings in the ModelType enum from the shared model module, which is needed for the type filtering functionality.
30-33: LGTM! Well-implemented model type filtering.The addition of the
typeprop to theModelSelectcomponent properly restricts the selectable models to Chat and ImageGeneration types, which is appropriate for the title view context. The implementation follows Vue prop syntax correctly and aligns with the broader UI enhancement for model filtering across components.src/renderer/src/i18n/ja-JP/update.json (1)
14-15: LGTM! Accurate Japanese translations added.The new localization keys
alreadyUpToDateandalreadyUpToDateDescprovide natural and appropriate Japanese translations for the "already up to date" state. The translations are accurate and consistent with the existing localization pattern.src/renderer/src/i18n/zh-CN/update.json (1)
14-15: LGTM! Accurate Chinese translations added.The new localization keys provide clear and natural Chinese translations for the "already up to date" state. The translations are appropriate for Chinese users and follow the established localization patterns.
src/renderer/src/i18n/fa-IR/update.json (1)
14-15: LGTM! Persian translations added consistently.The new localization keys follow the same pattern as other language files and provide appropriate Persian translations for the "already up to date" messaging. The translations appear contextually correct and maintain consistency with the established localization structure.
src/main/presenter/proxyConfig.ts (1)
149-150: Good cleanup: Removed unnecessary escape characters.The removal of escape characters before forward slashes in the regex pattern is correct. In JavaScript string literals, forward slashes don't need to be escaped, so this change improves readability without affecting functionality.
src/renderer/src/components/settings/ModelProviderSettingsDetail.vue (2)
41-41: Good reactive event handling.Adding the
config-changedevent listener ensures the component stays synchronized with configuration changes from the model manager.
348-352: Clean implementation of config change handler.The handler properly reinitializes data when configuration changes occur. The async/await pattern is correctly used with the existing
initData()function.src/main/presenter/mcpPresenter/inMemoryServers/filesystem.ts (1)
178-178: Acceptable removal of unused error parameter.The error parameter removal is appropriate here since the catch block is specifically handling the expected case where
fs.realpath()fails for new files that don't exist yet, and the error details aren't needed for the fallback logic.src/main/presenter/llmProviderPresenter/providers/anthropicProvider.ts (3)
240-242: Good formatting improvement.The improved indentation enhances readability of the chained method calls for filtering and mapping text content.
316-322: Consistent formatting for object literal.The improved indentation makes the image source object structure more readable and maintains consistency with the similar code block later in the file.
392-399: Consistent formatting applied.The formatting improvement matches the style used in the earlier similar code block, maintaining consistency throughout the file.
src/renderer/src/i18n/ru-RU/update.json (1)
14-15: LGTM!The Russian translations for the update status messages are properly formatted and follow the existing naming conventions. The content is appropriate for indicating that the application is already up to date.
src/main/presenter/windowPresenter/index.ts (1)
916-916: LGTM!Good fix for the camelCase naming convention. The method name
sendToDefaultTabis now properly formatted and consistent with standard naming practices.Verify that all call sites have been updated to use the corrected method name:
#!/bin/bash # Description: Search for any remaining calls to the old method name # Expected: Should only find the new corrected method name rg "sendTodefaultTab" --type ts --type js rg "sendToDefaultTab" --type ts --type js -A 2src/main/presenter/mcpPresenter/inMemoryServers/builder.ts (1)
34-34: LGTM!The addition of the
envparameter to the PowerpackServer constructor is consistent with other in-memory servers and enables E2B sandbox integration. This follows the established pattern used by other servers in the switch statement.src/renderer/src/components/settings/PromptSetting.vue (2)
183-470: LGTM!The formatting improvements enhance code readability and consistency. The changes include proper indentation, line breaks, and template expression formatting while maintaining the same functionality.
1005-1005: Good syntax improvement.Adding parentheses around the forEach callback parameter improves code consistency and follows JavaScript/TypeScript best practices.
src/main/presenter/threadPresenter/index.ts (1)
1969-1969: LGTM: Improved naming for function call record key.The renaming from
function_call_resulttofunction_call_recordimproves clarity and aligns with the enhanced tool call record handling mentioned in the AI summary.Please verify that this key name change is consistent across the codebase:
#!/bin/bash # Description: Check for any remaining usage of the old key name and verify consistency of the new key name. # Search for any remaining usage of the old key name echo "Searching for old key name 'function_call_result':" rg "function_call_result" --type ts --type js echo -e "\nSearching for new key name 'function_call_record':" rg "function_call_record" --type ts --type jssrc/main/presenter/threadPresenter/messageManager.ts (2)
99-104: LGTM: Improved code formatting for better readability.The multi-line formatting of the
eventBus.sendToRenderercall improves code readability without changing functionality.
286-286: LGTM: Removed unused exception variable.Good practice to remove the unused exception variable in the catch clause, which helps avoid linting warnings and keeps the code clean.
src/renderer/src/components/ui/UpdateDialog.vue (4)
5-5: LGTM: Dynamic dialog title improves user experience.The conditional title based on
upgrade.hasUpdateprovides clear context to users about whether an update is available or if they're already up to date.
8-28: LGTM: Conditional content rendering enhances dialog functionality.The conditional rendering properly displays:
- Update details (version, release date, notes, download progress) when an update is available
- "Already up to date" message when no update is found and not in silent mode
The download progress display with formatted size and speed information provides good user feedback during updates.
33-34: LGTM: Dynamic button labeling improves user interaction.The button label appropriately changes between "Later" and "Close" based on update availability, providing clearer context for the user action.
38-38: LGTM: Conditional install button display.The install button properly shows only when
upgrade.isReadyToInstallis true, preventing premature installation attempts.README.md (2)
266-267: LGTM! Package manager migration to pnpm.The commands have been correctly updated to use pnpm instead of npm, including the new
installRuntimescript. This aligns with the broader project migration to pnpm.
277-277: LGTM! Build commands updated for pnpm.All development and build commands have been correctly updated to use pnpm across all platforms and architectures.
Also applies to: 284-284, 287-287, 290-299
src/main/presenter/configPresenter/mcpConfHelper.ts (1)
251-255: LGTM! Improved event communication with explicit targeting.The change from
eventBus.emittoeventBus.sendwithSendTarget.ALL_WINDOWSimproves the reliability and clarity of event propagation to all renderer windows. This aligns with the broader refactoring of the event system mentioned in the AI summary.src/main/presenter/llmProviderPresenter/providers/deepseekProvider.ts (1)
80-81: LGTM! Code formatting improvements enhance readability.These formatting changes improve code consistency and readability:
- Header keys use unquoted identifiers (more common TypeScript style)
- Error message uses multi-line template string for better readability
- Balance info selection uses explicit parentheses and line breaks for clarity
No functional changes were introduced.
Also applies to: 87-89, 99-102
README.jp.md (3)
262-263: LGTM! Dependency commands updated to pnpm.The Japanese documentation correctly updates the dependency installation commands to use pnpm, maintaining consistency with the English version.
268-269: LGTM! Windows Developer Mode note added.The Japanese translation of the Windows-specific pnpm requirements is clear and helpful for Japanese users.
273-273: LGTM! Build commands updated across all platforms.All development and build commands have been correctly updated to use pnpm across all platforms and architectures in the Japanese documentation.
Also applies to: 280-280, 283-283, 286-286, 289-294
CONTRIBUTING.md (4)
23-23: LGTM! Development start command updated to pnpm.The basic functionality check instruction correctly uses
pnpm run devinstead of npm.
87-88: LGTM! Dependency installation commands updated.The installation instructions correctly use
pnpm installand include the newpnpm run installRuntimecommand, maintaining consistency with other documentation.
93-93: LGTM! Development server command updated.The development startup command correctly uses
pnpm run dev.
113-115: LGTM! Code style verification commands updated.All code quality commands (lint, build, i18n) have been correctly updated to use pnpm, ensuring contributors follow the proper development workflow.
src/renderer/src/components/settings/ProviderModelManager.vue (3)
39-39: LGTM: Provider ID prop addition enables provider-aware configuration.The addition of the
provider-idprop toModelConfigItemcorrectly enables tracking configuration changes at the provider level, which aligns with the enhanced model configuration management described in the PR objectives.
47-47: LGTM: Configuration change event propagation implemented correctly.The
config-changedevent emission fromModelConfigItemfollows Vue's event bubbling pattern and enables parent components to react to configuration updates.
74-74: LGTM: Event definition properly declared.The
config-changedevent is correctly added to the component's emit definitions, maintaining type safety in the component interface.src/renderer/src/components/settings/ProviderModelList.vue (3)
14-14: LGTM: Consistent provider-aware configuration for custom models.The addition of
provider-idprop andconfig-changedevent emission for custom models correctly implements the provider-aware configuration pattern.Also applies to: 23-23
111-111: LGTM: Consistent provider-aware configuration for provider models.The same pattern is correctly applied to provider models, ensuring uniform configuration management across both custom and official model sections.
Also applies to: 118-118
163-163: LGTM: Event definition maintains component interface consistency.The
config-changedevent is properly declared in the emit definitions, maintaining the same pattern as other components in this configuration management system..cursor/rules/development-setup.mdc (3)
4-4: LGTM: Configuration flag properly set.The
alwaysApply: trueflag ensures these development rules are consistently applied across the project.
7-11: LGTM: Simplified prerequisites align with project modernization.The concise list of prerequisites (OxLint, pnpm, English for logs/comments, Node.js >= 22, pnpm >= 9) reflects the project's migration to modern tooling as outlined in the PR objectives.
13-26: LGTM: Standardized pnpm commands streamline development workflow.The migration from npm to pnpm commands across all development tasks (install, lint, format, dev, build) is consistent with the project-wide package manager standardization mentioned in the PR summary.
README.zh.md (3)
262-263: LGTM: Installation commands updated for pnpm migration.The migration from
npm installtopnpm installand addition ofpnpm run installRuntimealigns with the project's standardization on pnpm as the package manager.
268-268: LGTM: Important Windows-specific guidance for pnpm.The note about enabling Developer Mode or using administrator accounts for Windows users is crucial for pnpm's symbolic link functionality. This prevents common installation failures on Windows systems.
273-273: LGTM: Consistent pnpm command migration across all development tasks.All development and build commands have been correctly updated from npm to pnpm, maintaining consistency with the project's package manager standardization.
Also applies to: 280-294
src/main/presenter/upgradePresenter/index.ts (2)
94-96: LGTM: Improved code formatting for event payload objects.The conversion from single-line to multi-line object literals improves code readability and maintainability without affecting functionality. This formatting style is more consistent with the project's coding standards.
Also applies to: 241-243, 306-308
327-332: LGTM: Consistent multi-line formatting for return object.The return object formatting matches the style applied to event payloads throughout the file, creating visual consistency and improving readability.
src/main/presenter/githubCopilotDeviceFlow.ts (4)
49-50: Improved error handling with logging.Good addition of error logging before throwing a generic error. This will help with debugging while maintaining a clean error interface for callers.
270-272: Better error handling clarity.Replacing the empty catch block with an explicit
// ignorecomment improves code readability and makes it clear that errors are intentionally being suppressed.
308-311: Good addition of polling safety limit.The polling count limit prevents potential infinite loops and provides a clear failure path. This is a solid improvement for robustness.
387-389: Consistent error handling approach.Same improvement as the earlier catch block - explicit ignore comment for clarity.
src/renderer/src/components/settings/ShortcutSettings.vue (2)
61-70: Simplified rendering with enhanced styling.The change from template v-for to direct span rendering is cleaner and the conditional styling for the '0' key (
font-mono tracking-widest) will improve visual distinction. The index-based keys are appropriate here since the array is stable.
160-163: New shortcut functionality added.The ShowHideWindow shortcut mapping follows the established pattern and uses an appropriate icon. This enhances the application's keyboard accessibility.
src/renderer/src/components/ModelSelect.vue (2)
63-68: Well-designed optional prop with proper typing.The
typeprop is properly typed withPropType<ModelType[]>and defaults toundefinedfor backward compatibility. The explicit default comment adds clarity.
111-116: Robust filtering implementation.The conditional filtering logic properly handles both the presence/absence of the
typeprop and empty arrays. The type checking withmodel.type !== undefinedprevents runtime errors.src/renderer/src/components/settings/CommonSettings.vue (2)
80-83: Appropriate model type restriction.Limiting the model selection to
ChatandImageGenerationtypes makes sense for a search assistant model configuration. This prevents users from selecting inappropriate model types.
489-489: Minor regex improvement.Removing the unnecessary escape character before the forward slash in the URL pattern is a small but correct cleanup.
src/main/presenter/mcpPresenter/index.ts (2)
224-227: Improved error logging readability.The multi-line error logging format is more readable and consistent with other error logging patterns in the codebase.
384-384: Consistent event emission pattern.Using
eventBus.sendwith explicitSendTarget.ALL_WINDOWStargeting aligns with the event dispatching pattern used elsewhere in the codebase and provides better control over event propagation.src/main/events.ts (2)
27-30: LGTM! Well-structured event constants for model configuration.The new model configuration events follow the established naming convention and provide comprehensive coverage for configuration changes, resets, and batch imports. These align well with the centralized model configuration management mentioned in the PR objectives.
158-158: Good addition for tray-based update functionality.The new
CHECK_FOR_UPDATESevent enables user-initiated update checks from the system tray, which enhances the user experience by providing easy access to update functionality..github/workflows/prcheck.yml (3)
23-28: Excellent migration to pnpm with proper version pinning.The migration from npm to pnpm is well-executed with specific version pinning for both Node.js (22.13.1) and pnpm (10.12.1), which ensures consistent builds across environments.
33-40: Good platform-specific configuration handling.The approach to configure pnpm workspace for different platforms and architectures is well-structured, with proper environment variable usage for TARGET_OS and TARGET_ARCH.
42-43: Great addition of lint checking.Adding the lint step to the PR check workflow will help maintain code quality and catch issues early in the development process.
src/main/presenter/tabPresenter.ts (2)
39-50: Excellent refactoring with centralized bounds update logic.The new
onWindowSizeChangemethod effectively centralizes the logic for updating tab view bounds when window dimensions change. This improves maintainability and reduces code duplication.
57-66: Smart timing approach for window state changes.The 100ms delay using
setTimeoutfor maximize/unmaximize events is a good practice that ensures the window state has fully settled before updating tab view bounds, preventing potential timing issues.src/renderer/src/components/settings/ProviderApiConfig.vue (1)
7-15: Great formatting improvements for better readability.The multi-line formatting of Vue component attributes significantly improves code readability and maintainability. These changes follow Vue.js best practices for template organization without altering any functionality.
Also applies to: 17-24, 35-40, 45-53, 55-64, 65-75, 77-82, 89-92, 180-183
src/renderer/src/i18n/ru-RU/settings.json (3)
104-160: Comprehensive model configuration localization.The extensive Russian localization for model configuration features is well-structured and comprehensive. It covers all aspects including labels, descriptions, validation messages, and dialog confirmations, properly supporting the new model configuration management features.
381-386: Good localization support for E2B sandbox integration.The Russian translations for the E2B sandbox feature are complete and informative, providing clear descriptions and help text for users. The validation message for required API key is appropriately included.
460-460: Appropriate shortcut localization addition.The Russian translation for the "showHideWindow" shortcut is correctly added and aligns with the new shortcut functionality mentioned in the PR objectives.
src/renderer/src/components/NewThread.vue (2)
61-64: LGTM: Model type filtering implementationThe addition of the
typeprop to restrict model selection to Chat and ImageGeneration models is a good enhancement that aligns with the broader model filtering improvements across the application.
120-120: LGTM: ModelType importThe import of
ModelTypeenum is correctly added to support the new model type filtering functionality.src/main/presenter/llmProviderPresenter/providers/openAICompatibleProvider.ts (2)
53-53: LGTM: Improved provider extensibilityChanging
isNoModelsApifromprivatetoprotectedallows subclasses to access this property, enhancing the extensibility of the OpenAI-compatible provider pattern. This aligns well with the introduction of new providers like MinimaxProvider and TogetherProvider.
1117-1121: LGTM: Enhanced function call ID generationThe improved fallback ID generation logic creates more descriptive and unique IDs by incorporating the function name, index, and timestamp. This enhancement will significantly improve debugging and identification of function calls when
parsedCall.idis missing.The format
${functionName}-${index}-${Date.now()}provides:
- Function context for easier debugging
- Index for ordering within the same response
- Timestamp for global uniqueness
src/renderer/src/components/settings/ModelConfigItem.vue (3)
42-50: LGTM: Well-implemented configuration buttonThe "Configure Model" button is properly implemented with:
- Consistent styling matching other action buttons
- Accessible title attribute using internationalization
- Clear visual representation with settings icon
63-70: LGTM: Clean dialog integrationThe
ModelConfigDialogintegration follows Vue 3 best practices with:
- Proper v-model binding for open state
- All necessary props passed down
- Event handling for configuration changes
84-84: LGTM: Proper prop and event handlingThe addition of
providerIdprop andconfigChangedevent provides the necessary interface for parent components to:
- Pass provider context to the configuration dialog
- React to configuration updates by refreshing data
The event emission in
onConfigSavedensures parent components can stay synchronized with configuration changes.Also applies to: 101-101, 104-114
.github/workflows/build.yml (4)
35-35: LGTM: Consistent Node.js version pinningPinning Node.js to version 22.13.1 across all platforms ensures consistent build environments and reduces potential version-related issues.
Also applies to: 92-92, 146-146
37-40: LGTM: Proper pnpm setupThe pnpm setup with version 10.12.1 is correctly configured across all platforms, ensuring consistent package management behavior.
Also applies to: 94-97, 148-151
45-50: LGTM: Platform-specific sharp configurationThe platform-specific sharp installation configuration is well-designed:
- Uses environment variables to specify target OS and architecture
- Handles native dependency compilation correctly
- Maintains separation of concerns between workspace configuration and dependency installation
Also applies to: 102-107, 156-161
180-180: LGTM: macOS memory optimizationAdding
NODE_OPTIONS: '--max-old-space-size=4096'for macOS builds is a good practice to prevent out-of-memory errors during the build process, especially for resource-intensive operations.scripts/install-sharp-for-platform.js (1)
1-96: LGTM! Well-structured platform configuration script.The script is well-implemented with:
- Proper environment variable handling with fallbacks
- Comprehensive platform support
- Good error handling and user feedback
- Atomic file operations with proper YAML formatting
The approach of dynamically updating the workspace configuration based on target platform is sound and integrates well with the pnpm migration.
src/main/presenter/configPresenter/providerModelSettings.ts (2)
106-174: LGTM! Comprehensive Gemini model configuration updates.The updates to Gemini models are well-structured:
- Model IDs and names are properly updated to reflect new versions
- Context lengths and token limits are reasonable for the respective models
- The addition of reasoning capabilities for newer models is appropriate
- The new
ImageGenerationmodel type forgemini-2.0-flash-preview-image-generationis properly configuredThe parameter values (context lengths up to 1M+ tokens, reasoning enabled for 2.x models) align with Google's documented Gemini capabilities.
460-470: LGTM! Proper addition of Minimax M1 80K model.The new PPIO model configuration is correctly implemented:
- Reasonable parameters for a high-context model (128K context, 40K max tokens)
- Appropriate capabilities (function calls and reasoning enabled)
- Proper match pattern for model ID resolution
src/main/presenter/syncPresenter/index.ts (4)
28-28: LGTM! Proper constant definition for model config path.The addition of
MODEL_CONFIG_PATHfollows the established pattern and uses the correct file location in userData directory.
167-171: LGTM! Complete model config backup and restore implementation.The model config file handling is properly integrated:
- Backup creation during import (lines 167-171)
- Restore from backup during import (lines 226-230)
- Rollback on failure (lines 256-260)
The implementation follows the same patterns as other config files and includes proper error handling.
Also applies to: 226-230, 256-260
331-341: LGTM! Comprehensive backup process integration.The model config file is properly integrated into the backup process:
- Temporary backup file creation with timestamp
- Atomic file replacement to prevent corruption
- Proper cleanup of old backup files before creating new ones
The implementation is consistent with other config files and maintains data integrity.
Also applies to: 381-439
285-288: LGTM! Proper cleanup of temporary model config files.The temporary file cleanup includes the model config file, ensuring no orphaned temporary files remain after import operations.
src/renderer/src/i18n/en-US/settings.json (3)
104-160: LGTM! Comprehensive model configuration localization.The new
modelConfigsection provides excellent UI localization:
- Clear labels and descriptions for all model parameters
- Comprehensive validation messages with specific limits
- User-friendly confirmation dialogs for destructive actions
- Proper categorization of model types
The text is clear, professional, and provides sufficient guidance for users configuring model parameters.
389-394: LGTM! Complete E2B sandbox localization.The E2B integration strings are well-structured:
- Clear feature description and purpose
- Helpful placeholder text and guidance
- Proper validation messaging
- Link to documentation for API key acquisition
460-460: LGTM! Appropriate shortcut key description.The new
showHideWindowshortcut description is clear and follows the established pattern for other shortcut descriptions.src/renderer/src/i18n/ja-JP/settings.json (3)
104-160: LGTM! Complete Japanese model configuration localization.The Japanese translation of the
modelConfigsection maintains the same comprehensive structure as the English version:
- All model parameter labels and descriptions are translated
- Validation messages with proper numeric limits
- Confirmation dialogs and button text
- Model type options properly localized
The structure is consistent and complete.
381-386: LGTM! Complete E2B sandbox Japanese localization.The E2B feature localization is properly translated:
- Feature descriptions and help text
- API key field labels and placeholders
- Validation requirements
- External documentation references
460-460: LGTM! Consistent shortcut key translation.The Japanese translation of the
showHideWindowshortcut description is appropriately localized and consistent with the pattern.src/renderer/src/components/settings/ModelConfigDialog.vue (2)
1-151: LGTM! Well-structured Vue component with comprehensive model configuration UI.The component follows Vue best practices with proper reactive data management, form validation, and internationalization. The template structure is clean and accessible with proper semantic HTML.
268-268: Let’s search the entire repo (without relying on unsupported--type vue) for both the hard-coded100_000_000limit and allcontextLengthreferences:#!/bin/bash # Search for the magic number across .vue and .ts files rg -g "*.vue" -g "*.ts" -n "100_000_000" -A3 -B3 # Search for all contextLength usage to see if defaults or other limits exist rg -g "*.vue" -g "*.ts" -n "contextLength" -A3 -B3src/main/presenter/mcpPresenter/serverManager.ts (4)
20-20: Good addition of uvRegistry property.The new private property follows the existing pattern and naming conventions in the class.
85-91: Conditional uvRegistry assignment is well-implemented.The logic correctly sets the Tsinghua mirror URL when using the npmmirror registry, providing a Chinese-specific UV registry for improved performance in that region.
100-104: Clean getter implementation.The getter method follows the same pattern as the existing
getNpmRegistry()method and maintains consistency.
183-188: Verify McpClient constructor signature change.The code passes
uvRegistryas a fourth parameter to the McpClient constructor. Ensure that the McpClient class has been updated to accept this parameter.#!/bin/bash # Check McpClient constructor to verify it accepts the uvRegistry parameter ast-grep --pattern $'class McpClient { $$$ constructor($_, $_, $_, $_) { $$$ } $$$ }'src/renderer/src/i18n/zh-CN/settings.json (3)
104-160: Comprehensive and well-structured model configuration translations.The Chinese translations for the model configuration section are thorough, covering all UI elements, validation messages, and user interactions. The translation quality appears accurate and user-friendly.
387-394: Clear E2B sandbox translations with helpful guidance.The E2B sandbox translations provide clear descriptions and helpful guidance for users, including proper placeholders and validation messages.
460-460: Appropriate shortcut translation.The "显示/隐藏窗口" translation for the showHideWindow shortcut is clear and concise.
src/renderer/src/i18n/ko-KR/settings.json (3)
104-160: Well-translated Korean model configuration section.The Korean translations maintain the same structure as the Chinese version and provide clear, user-friendly text for all model configuration elements.
380-387: Clear Korean E2B sandbox translations.The E2B sandbox feature translations are well-localized with appropriate Korean terminology and helpful user guidance.
460-460: Appropriate Korean shortcut translation.The "창을 표시/숨기기" translation correctly conveys the show/hide window functionality.
src/renderer/src/i18n/fa-IR/settings.json (3)
104-160: Comprehensive Persian model configuration translations.The Persian translations provide complete coverage of the model configuration interface with appropriate terminology. The structure matches other language files for consistency.
388-395: Well-localized Persian E2B sandbox translations.The E2B sandbox feature translations use appropriate Persian terminology and provide clear user guidance for the sandbox functionality.
460-460: Appropriate Persian shortcut translation.The "نمایش/مخفی کردن پنجره" translation accurately represents the show/hide window functionality in Persian.
src/renderer/src/i18n/fr-FR/settings.json (3)
104-160: LGTM! Comprehensive French translations for model configuration.The new
modelConfigsection provides thorough French localization for the model configuration interface, including all necessary UI elements, validation messages, and descriptions. The translations are consistent and appropriately localized.
381-386: LGTM! E2B sandbox localization is complete.The French translations for E2B sandbox functionality are well-structured and provide clear guidance for users, including proper help text and validation messages.
460-460: LGTM! New shortcut translation added.The translation for the new "showHideWindow" shortcut is appropriate and consistent with other shortcut translations.
src/main/presenter/llmProviderPresenter/providers/minimaxProvider.ts (2)
11-36: LGTM! Model metadata definition is comprehensive.The hardcoded model metadata provides appropriate default configurations for MiniMax models with consistent parameter values.
38-84: LGTM! Method implementations are clean and consistent.All three methods properly delegate to the shared
openAICompletionfunction with appropriate message formatting.src/main/presenter/llmProviderPresenter/providers/togetherProvider.ts (2)
65-82: Verify fixed model parameters for accuracy.The implementation uses fixed
contextLength: 4096andmaxTokens: 2048for all Together AI models, which may not accurately represent the actual capabilities of different models.Consider using the actual model specifications from the Together AI API response if available, or verify these values are appropriate defaults for all Together AI models.
56-63: LGTM! Dynamic model fetching implementation is well-structured.The dynamic model fetching using the Together SDK with appropriate filtering for chat/language models is a good approach.
src/main/presenter/configPresenter/modelConfig.ts (5)
9-16: LGTM! Clean initialization with appropriate store naming.The constructor properly initializes the electron-store with a descriptive name for model configurations.
24-67: LGTM! Excellent priority-based configuration resolution.The
getModelConfigmethod implements a logical priority system:
- User-defined config (highest priority)
- Provider-specific defaults
- Pattern-matched defaults
- Safe fallback (lowest priority)
This approach provides flexibility while ensuring the system always has valid configuration.
75-81: LGTM! Clean configuration persistence.The
setModelConfigmethod properly stores configurations with a clear key structure using the special concatenation character.
138-150: LGTM! Robust import functionality with overwrite control.The import method provides both incremental and overwrite modes, which is essential for synchronization and backup restoration scenarios.
105-120: LGTM! Efficient provider-specific configuration retrieval.The method efficiently filters configurations by provider using the concatenation character as a delimiter.
src/main/presenter/configPresenter/modelDefaultSettings.ts (5)
340-382: LGTM! Gemini model standardization is well-executed.The updates to Gemini models provide consistent naming, standardized context lengths (1M tokens), increased max tokens (65536), and proper reasoning flags. The model ID updates align with Google's API patterns.
395-408: LGTM! Explicit model type for image generation.The addition of
type: ModelType.ImageGenerationfor the Gemini 2.0 Flash Preview Image Generation model properly categorizes this specialized model.
1345-1355: LGTM! New MiniMax model configuration is comprehensive.The new MiniMax M1 80K model entry includes appropriate parameters with function call and reasoning support enabled.
1672-1684: LGTM! Embedding model type specification is correct.The addition of explicit
ModelType.Embeddingfor embedding models properly categorizes these specialized models for UI filtering and functionality.
343-356: Verify the large context lengths are accurate.Several Gemini models specify very large context lengths (1,048,576 tokens). Please verify these values match the actual model capabilities from Google's documentation.
What are the actual context length limits for Gemini 2.5 Pro, Gemini 2.5 Flash, and Gemini 2.0 Flash models according to Google's latest documentation?src/main/presenter/llmProviderPresenter/providers/ppioProvider.ts (2)
10-23: LGTM! Well-structured interface for PPIO API response.The interface properly defines all necessary fields for model metadata with appropriate types.
89-89: ```shell
#!/bin/bashInspect the ppioProvider implementation around the authorization header
FILE=src/main/presenter/llmProviderPresenter/providers/ppioProvider.ts
if [ -f "$FILE" ]; then
echo "Inspecting $FILE (lines 80–100):"
sed -n '80,100p' "$FILE"
else
echo "File not found: $FILE"
fi</details> <details> <summary>package.json (3)</summary> `1-12`: **Version update and package manager migration look good.** The migration to pnpm with proper version constraints and package manager locking is well implemented. Node.js 20.12.2+ requirement ensures compatibility with modern features. --- `13-54`: **Script updates are well-structured for pnpm migration.** The consistent replacement of npm with pnpm, addition of Vitest testing commands, and runtime management improvements are all properly implemented. --- `56-90`: **New dependencies align with PR objectives.** The addition of @e2b/code-interpreter for sandbox support and together-ai for the new provider are appropriate. The minor version updates for existing packages look safe. </details> <details> <summary>src/main/presenter/llmProviderPresenter/baseProvider.ts (2)</summary> `243-244`: **Excellent security enhancement for tool call handling.** The explicit rules preventing exposure of function_call XML tags and structures help prevent potential prompt injection or information leakage attacks. --- `257-337`: **Clear improvements to tool call documentation and constraints.** The reorganized constraints with clearer explanations about external system insertion of tool records, response format flexibility (JSON or string), and explicit language handling improve the robustness of the prompt system. </details> <details> <summary>src/main/presenter/llmProviderPresenter/index.ts (3)</summary> `29-36`: **New provider integrations look good.** The MinimaxProvider and TogetherProvider are properly imported and integrated into the provider creation logic following the established pattern. Also applies to: 95-97, 146-147 --- `200-209`: **Robust error handling for provider initialization.** The sequential initialization with try-catch blocks ensures that a single provider failure doesn't prevent other providers from initializing. Good defensive programming. --- `1031-1042`: **Good defensive programming with fallback mechanism.** The try-catch with fallback to configPresenter ensures that custom models can still be retrieved even if the provider instance isn't properly initialized. The warning log provides useful debugging information. </details> <details> <summary>src/main/presenter/llmProviderPresenter/providers/openRouterProvider.ts (2)</summary> `21-52`: **Well-structured interface definition** The `OpenRouterModelResponse` interface comprehensively captures the OpenRouter API response structure with proper typing and optional field markers. --- `117-119`: **Clean formatting improvements** The formatting changes improve code readability and error message clarity. Also applies to: 124-127, 141-141, 146-146 </details> <details> <summary>src/renderer/src/components/mcp-config/mcpServerForm.vue (5)</summary> `66-69`: **Proper E2B state management and validation** The E2B configuration state is correctly initialized and validated, ensuring the API key is required when E2B is enabled. Also applies to: 283-288 --- `833-834`: **Enhanced model type filtering** Supporting multiple model types in the selection component improves flexibility and aligns with the new model categorization system. --- `462-470`: **Secure E2B configuration handling** The E2B configuration is properly injected into environment variables with appropriate conditional logic for the API key. --- `596-602`: **Robust E2B configuration parsing** The configuration parsing correctly handles type conversion and is properly scoped to the powerpack server. --- `967-970`: **Secure API key handling in UI** Good security practice using password input type for the E2B API key field. </details> <details> <summary>src/main/presenter/configPresenter/index.ts (4)</summary> `172-181`: **Correct data migration for minimax provider** The migration properly fixes the minimax baseUrl for older versions and uses proper version comparison. --- `322-332`: **Efficient batch model status retrieval** The batch retrieval method improves performance by reducing multiple store operations to a single batch operation. --- `428-434`: **Performance improvement in getAllEnabledModels** Excellent use of batch retrieval to optimize the filtering of enabled models. --- `839-849`: **Clean delegation pattern with proper event handling** The model configuration methods correctly delegate to the helper class and emit appropriate events to keep all windows synchronized. Also applies to: 856-865, 903-906 </details> <details> <summary>src/main/presenter/mcpPresenter/inMemoryServers/powerpackServer.ts (5)</summary> `120-132`: **Robust E2B configuration validation** The configuration parsing correctly validates the API key presence and provides appropriate warnings when misconfigured. --- `134-176`: **Comprehensive runtime detection** The unified runtime setup correctly handles platform-specific paths and provides clear logging about available runtimes. --- `286-344`: **Secure E2B sandbox execution with proper cleanup** The E2B execution implementation correctly handles sandbox lifecycle with guaranteed cleanup in the finally block, preventing resource leaks. --- `370-410`: **Clear separation of execution modes** The tool exposure logic properly separates E2B and local execution modes, preventing mixing of execution environments. --- `506-508`: **Proper enforcement of E2B-exclusive execution** Good security practice disabling local code execution when E2B is enabled, ensuring consistent sandbox isolation. Also applies to: 534-536 </details> <details> <summary>src/main/presenter/llmProviderPresenter/providers/geminiProvider.ts (4)</summary> `51-137`: **Comprehensive static model list with latest Gemini models** The updated model list includes all current Gemini models with accurate capabilities and proper typing for the image generation model. --- `309-350`: **Well-designed auto-enable logic respecting user preferences** The implementation correctly auto-enables only recommended models while respecting any existing manual configurations or custom models. --- `444-449`: **Improved model type detection** Good refactoring to use dynamic model type detection instead of hardcoded model IDs, improving maintainability. Also applies to: 873-873 --- `188-200`: **Consider maintaining capability detection heuristics** The capability detection relies on model name patterns which may become outdated as new models are released. Consider implementing a more maintainable approach or documenting these heuristics for regular updates. ```web What are the latest Gemini model capabilities and naming patterns?src/main/presenter/mcpPresenter/mcpClient.ts (4)
6-6: LGTM! Event dispatching pattern update.The change from
eventBus.emittoeventBus.sendwithSendTargetis consistent with the broader refactoring across MCP-related modules.
47-84: Well-designed session error detection mechanism.The session error detection logic is comprehensive with appropriate pattern matching for both session-specific and HTTP-related errors. The distinction between high and low confidence patterns is a good approach.
227-237: Correct implementation of UV registry support.The UV registry handling properly adds the
--indexparameter when a registry is configured, maintaining compatibility with UV's command-line interface.
278-337: Comprehensive runtime detection logic.The runtime path initialization properly checks for executable existence across different platforms and gracefully handles missing runtimes.
| // 托盘 检测更新 | ||
| eventBus.on(TRAY_EVENTS.CHECK_FOR_UPDATES, () => { | ||
| const allWindows = presenter.windowPresenter.getAllWindows() | ||
|
|
||
| // 查找目标窗口 (焦点窗口或第一个窗口) | ||
| const targetWindow = presenter.windowPresenter.getFocusedWindow() || allWindows![0] | ||
| presenter.windowPresenter.show(targetWindow.id) | ||
| targetWindow.focus() // 确保窗口置顶 | ||
|
|
||
| // 触发更新 | ||
| presenter.upgradePresenter.checkUpdate() | ||
| }) |
There was a problem hiding this comment.
Good implementation, but fix potential null reference issue.
The event listener logic correctly handles window management and update checking. However, there's a potential issue on line 98.
Apply this diff to fix the potential null reference:
- const targetWindow = presenter.windowPresenter.getFocusedWindow() || allWindows![0]
+ const targetWindow = presenter.windowPresenter.getFocusedWindow() || allWindows[0]The non-null assertion operator (!) is unnecessary and could cause runtime errors if getAllWindows() returns null or undefined.
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| // 托盘 检测更新 | |
| eventBus.on(TRAY_EVENTS.CHECK_FOR_UPDATES, () => { | |
| const allWindows = presenter.windowPresenter.getAllWindows() | |
| // 查找目标窗口 (焦点窗口或第一个窗口) | |
| const targetWindow = presenter.windowPresenter.getFocusedWindow() || allWindows![0] | |
| presenter.windowPresenter.show(targetWindow.id) | |
| targetWindow.focus() // 确保窗口置顶 | |
| // 触发更新 | |
| presenter.upgradePresenter.checkUpdate() | |
| }) | |
| // 托盘 检测更新 | |
| eventBus.on(TRAY_EVENTS.CHECK_FOR_UPDATES, () => { | |
| const allWindows = presenter.windowPresenter.getAllWindows() | |
| // 查找目标窗口 (焦点窗口或第一个窗口) | |
| const targetWindow = presenter.windowPresenter.getFocusedWindow() || allWindows[0] | |
| presenter.windowPresenter.show(targetWindow.id) | |
| targetWindow.focus() // 确保窗口置顶 | |
| // 触发更新 | |
| presenter.upgradePresenter.checkUpdate() | |
| }) |
🤖 Prompt for AI Agents
In src/main/index.ts around lines 93 to 104, remove the non-null assertion
operator (!) after getAllWindows() on line 98 to avoid potential runtime errors
if getAllWindows() returns null or undefined. Instead, add a check to ensure
allWindows is not null or undefined before accessing its first element, and
handle the case where no windows are available to prevent null reference
exceptions.
| } catch { | ||
| // 事务会自动回滚，直接抛出错误 | ||
| throw error | ||
| throw new Error('Failed to import data') |
There was a problem hiding this comment.
🛠️ Refactor suggestion
Consider preserving original error information.
The change removes the original error details by throwing a generic error message instead of re-throwing the caught error. This obscures valuable debugging information.
Consider preserving the original error context:
- } catch {
- // 事务会自动回滚，直接抛出错误
- throw new Error('Failed to import data')
+ } catch (error) {
+ // 事务会自动回滚，直接抛出错误
+ throw new Error(`Failed to import data: ${error instanceof Error ? error.message : String(error)}`)📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| } catch { | |
| // 事务会自动回滚，直接抛出错误 | |
| throw error | |
| throw new Error('Failed to import data') | |
| } catch (error) { | |
| // 事务会自动回滚，直接抛出错误 | |
| throw new Error(`Failed to import data: ${error instanceof Error ? error.message : String(error)}`) | |
| } |
🤖 Prompt for AI Agents
In src/main/presenter/sqlitePresenter/importData.ts around lines 102 to 104, the
catch block throws a new generic error, losing the original error details.
Modify the catch block to capture the original error object and either re-throw
it directly or include its message and stack trace in the new error to preserve
debugging information.
| run: pnpm run i18n | ||
|
|
||
| - name: Build | ||
| run: pnpm run build |
There was a problem hiding this comment.
Fix missing newline at end of file.
The file is missing a newline character at the end, which violates YAML formatting standards.
- run: pnpm run build
+ run: pnpm run build
+📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| run: pnpm run build | |
| run: pnpm run build | |
🧰 Tools
🪛 YAMLlint (1.37.1)
[error] 49-49: no new line character at the end of file
(new-line-at-end-of-file)
🤖 Prompt for AI Agents
In .github/workflows/prcheck.yml at line 49, add a newline character at the end
of the file to comply with YAML formatting standards and avoid potential parsing
issues.
| async summaries( | ||
| text: string, | ||
| modelId: string, | ||
| temperature?: number, | ||
| maxTokens?: number | ||
| ): Promise<LLMResponse> { | ||
| return this.openAICompletion( | ||
| [ | ||
| { | ||
| role: 'user', | ||
| content: `请总结以下内容，使用简洁的语言，突出重点：\n${text}` | ||
| } | ||
| ], | ||
| modelId, | ||
| temperature, | ||
| maxTokens | ||
| ) | ||
| } |
There was a problem hiding this comment.
🛠️ Refactor suggestion
Fix language inconsistency in summary prompt.
The summary prompt is in Chinese while the MinimaxProvider uses English. This inconsistency should be addressed for a cohesive user experience.
Consider standardizing the language or making it configurable:
- content: `请总结以下内容，使用简洁的语言，突出重点：\n${text}`
+ content: `Please summarize the following content using concise language and highlighting key points:\n${text}`📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| async summaries( | |
| text: string, | |
| modelId: string, | |
| temperature?: number, | |
| maxTokens?: number | |
| ): Promise<LLMResponse> { | |
| return this.openAICompletion( | |
| [ | |
| { | |
| role: 'user', | |
| content: `请总结以下内容，使用简洁的语言，突出重点：\n${text}` | |
| } | |
| ], | |
| modelId, | |
| temperature, | |
| maxTokens | |
| ) | |
| } | |
| async summaries( | |
| text: string, | |
| modelId: string, | |
| temperature?: number, | |
| maxTokens?: number | |
| ): Promise<LLMResponse> { | |
| return this.openAICompletion( | |
| [ | |
| { | |
| role: 'user', | |
| content: `Please summarize the following content using concise language and highlighting key points:\n${text}` | |
| } | |
| ], | |
| modelId, | |
| temperature, | |
| maxTokens | |
| ) | |
| } |
🤖 Prompt for AI Agents
In src/main/presenter/llmProviderPresenter/providers/togetherProvider.ts between
lines 19 and 36, the summary prompt is currently hardcoded in Chinese, causing
inconsistency with other providers like MinimaxProvider that use English. To fix
this, update the prompt text to English or make the prompt language configurable
via a parameter or configuration setting, ensuring consistent language usage
across providers for a cohesive user experience.
Release 0.2.4
Summary by CodeRabbit
New Features
Improvements
Bug Fixes
Documentation
Chores