- Updated globalMutate calls to use createKeyMatcher and createMultiKeyMatcher for improved cache key handling.
- Modified return types in buildSWRKey to support optional query parameters.
- Added createKeyMatcher and createMultiKeyMatcher functions to facilitate matching of SWR cache keys by path and multiple paths, respectively.
- Introduced methods to retrieve comprehensive statistics for preferences, including summary and detailed per-key information.
- Implemented a new `getStats` method to provide insights into subscription counts and active windows.
- Enhanced the PreferenceService with additional interfaces for better type safety and clarity in statistics handling.
- Updated documentation to reflect the new statistics features and their usage.
- Implemented comprehensive cache statistics in CacheService, allowing retrieval of summary and detailed information about cache tiers (memory, shared, persist).
- Introduced methods to estimate memory usage and format byte sizes for better readability.
- Enhanced the Inputbar component to log cache statistics for monitoring purposes.
- Updated related types to support new cache statistics features.
- Updated package.json to remove redundant i18n:check command.
- Added a "careers" section in the AboutSettings component with a button linking to the careers page.
- Introduced translations for the "careers" section in multiple languages including English, Chinese, German, Spanish, French, Japanese, Portuguese, Romanian, and Russian.
- Updated cache-related translations across various languages to provide localized support.
* ci(workflows): add pnpm caching to improve build performance
Cache pnpm dependencies to reduce installation time in CI workflows
* ci(github-actions): change claude-code-action version to v1
* ci(workflows): improve issue tracker formatting
Format issue body as markdown code block for better readability in workflow output
* ci(workflows): unify claude-code-action version to v1 for both jobs
Changed process-pending-issues job from @main to @v1 to maintain consistency across all jobs and avoid potential version conflicts.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(workflows): remove --help from allowed-tools to avoid upstream parsing bug
The `--help` flags in claude_args were being incorrectly parsed as actual
command-line arguments by claude-code-action, causing JSON parsing errors
("Unexpected identifier 'Usage'").
Switched to wildcard pattern `pnpm tsx scripts/feishu-notify.ts*` which:
- Allows all feishu-notify.ts commands including --help
- Avoids triggering the upstream argument parsing bug
- Simplifies the allowed-tools configuration
This addresses the root cause identified by Anthropic engineers.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(workflows): remove --help references from prompts
Remove `--help` references from prompts to avoid potential parsing issues.
The example commands are already comprehensive enough without needing to
mention the help flag.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(security): prevent path traversal vulnerability in DXT plugin system
Add input validation to prevent path traversal attacks in DXT plugin handling:
- Add sanitizeName() to filter dangerous characters from manifest.name
- Add validateCommand() to reject commands with path traversal sequences
- Add validateArgs() to validate command arguments
- Remove unsafe fallback logic in cleanupDxtServer()
The vulnerability allowed attackers to write files to arbitrary locations
on Windows by crafting malicious DXT packages with path traversal sequences
(e.g., "..\\..\\Windows\\System32\\") in manifest.name or command fields.
* refactor: use path validation instead of input sanitization
---------
Co-authored-by: defi-failure <159208748+defi-failure@users.noreply.github.com>
- Refactored the StreamingService to replace all instances of "session" with "task" to better reflect the functionality and improve clarity.
- Updated related types, methods, and cache handling to align with the new terminology, ensuring consistent usage throughout the codebase.
- Enhanced comments and documentation to reflect the changes in terminology and functionality.
sendReasoning is a patch specific to @ai-sdk/openai-compatible package.
Previously it was incorrectly applied to all providers in buildGenericProviderOptions,
including those with dedicated SDK packages (e.g., cerebras, deepseek, openrouter).
Now it only applies when the provider will actually use openai-compatible SDK:
- No dedicated SDK registered (!hasProviderConfig(providerId))
- OR explicitly openai-compatible (providerId === 'openai-compatible')
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
- Updated the StreamingService to create new immutable instances of sessions and messages when adding blocks or applying updates. This change enhances cache notification mechanisms by ensuring that references are updated correctly.
- Improved the handling of block additions and message updates to maintain immutability, which aids in better state management and performance.
* ci(workflows): add feishu notification for failed sync jobs
Add Feishu webhook notification when sync-to-gitcode workflow fails or is cancelled. The notification includes tag name, status and run URL for quick debugging.
* ci(workflow): add feishu notification for failed or cancelled jobs
* fix: add dispose method to prevent abort listener leak
Add dispose() method to StreamAbortController that explicitly removes
the abort event listener when stream ends normally. Previously, the
listener would only be removed when abort was triggered ({ once: true }),
but if the stream completed normally without abort, the listener would
remain attached until garbage collection.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* chore: format code
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
Updated version number to 1.7.12 in package.json and electron-builder.yml. Added release notes detailing the introduction of the MCP Hub with Auto mode and new cache control options for the Anthropic provider, along with various bug fixes.
fix: set X11 window class and name to cherry-studio
Set window class and name for Linux X11 to ensure system tray and
window manager identify the app correctly instead of using default
'electron' identifier.
* refactor: use pnpm install instead of manual download for prebuild packages
Replace manual tgz download with pnpm install for architecture-specific
prebuild binaries, simplifying the build process.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* delete utils
* update after pack
* udpate before pack
* use optional deps
* refactor: use js-yaml to modify pnpm-workspace.yaml for cross-platform builds
- Add all prebuild packages to optionalDependencies in package.json
- Use js-yaml to parse and modify pnpm-workspace.yaml
- Add target platform to supportedArchitectures.os and cpu
- Restore original config after pnpm install
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix version
* refactor: streamline package management and filtering logic in before… (#12370)
refactor: streamline package management and filtering logic in before-pack.js
- Consolidated architecture-specific package definitions into a single array for better maintainability.
- Simplified the logic for determining target platform and architecture.
- Enhanced the filtering process for excluding and including packages based on architecture and platform.
- Improved console logging for clarity during package installation.
This refactor aims to improve the readability and efficiency of the prebuild package handling process.
* refactor: update package filtering logic in before-pack.js to read from electron-builder.yml
- Modified the package filtering process to load configuration directly from electron-builder.yml, reducing potential errors from multiple overrides.
- Enhanced maintainability by centralizing the file configuration management.
This change aims to streamline the prebuild package handling and improve configuration clarity.
---------
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
Co-authored-by: 亢奋猫 <kangfenmao@qq.com>
* build: add commander package as dependency
* refactor(scripts): migrate feishu-notify to TypeScript with CLI interface
- Convert JavaScript implementation to TypeScript with proper type definitions
- Add CLI interface using commander for better usability
- Improve error handling and input validation
- Add version management and subcommand support
* ci(workflows): update feishu notification command and add pnpm install step
Update the feishu notification command to use CLI tool with proper arguments instead of direct node script execution
Add pnpm install step to ensure dependencies are available before running the workflow
* docs: add feishu notification script documentation
Add Chinese and English documentation for the feishu-notify.ts CLI tool
* feat(notify): add generic send command to feishu-notify
Add a new 'send' subcommand to send simple notifications to Feishu with customizable title, description and header color. This provides a more flexible way to send notifications without being tied to specific business logic like the existing 'issue' command.
The implementation includes:
- New send command handler and options interface
- Simple card creation function
- Zod schema for header color validation
- Documentation updates in both Chinese and English
- Enhanced the logic to determine included Claude code vendors based on architecture and platform.
- Adjusted filters for excluding and including Claude code vendors to improve compatibility, particularly for Windows ARM64.
- Removed unnecessary variables and streamlined the filter application process.
- Removed outdated dependencies: js-yaml, bonjour-service, and emoji-picker-element-data from devDependencies.
- Added js-yaml, bonjour-service, and emoji-picker-element-data back to their respective locations in dependencies.
- Introduced optionalDependencies for @strongtz/win32-arm64-msvc with version ^0.4.7.
- Updated pnpm-lock.yaml to reflect the changes in package.json and added new package versions.
- Documented key architectural concerns in the types index file, including file size, responsibility separation, and organization of utility functions and constants.
- Suggested migration of certain types to shared packages for better maintainability and clarity.
- Added new section on schema file organization, detailing principles and decision criteria for merging or separating table files.
- Updated file naming conventions for single and multi-table files, as well as helper utilities.
- Refactored import paths in various schema files to use the new `_columnHelpers` module instead of the deprecated `columnHelpers`.
- Removed obsolete `customSql.ts`, `columnHelpers.ts`, `messageFts.ts`, `tag.ts`, `entityTag.ts`, and other related files to streamline the codebase.
This refactor improves clarity in database schema management and aligns file organization with best practices.
Updated version number to 1.7.11 in package.json and electron-builder.yml. Added release notes highlighting the introduction of the MCP Hub with Auto mode and various bug fixes, including improvements to the Chat and Editor components.
This commit fixes a bug where search results could not be retrieved from Bing, Baidu, and Google.
The root cause of this issue was a discrepancy in page content when the Electron window was hidden versus when it was visible. Additionally, the previous use of `did-finish-load` caused page jitter within the window, leading to sporadic failures in fetching search content.
To resolve this, I've enabled offscreen rendering, ensuring consistent page content regardless of window visibility. Furthermore, I've switched to using the `ready-to-show` event to ensure the complete page DOM is available before attempting to retrieve content, thereby eliminating the search bug.
* feat(fetch): add request throttling (already present in the original, keeping it)
Co-authored-by: suyao <sy20010504@gmail.com>
* refactor: separate message extraction from rendering
Extract `lastAssistantMessage` memoization separately from rendering
`MessageContent` component, improving code clarity and separation of
concerns.
* feat: Replace manual loading state with AssistantMessageStatus tracking
* refactor: Replace loading state with status enum in translation action
- Add LoadingOutlined icon for preparing state
- Remove AssistantMessageStatus dependency
- Simplify streaming detection using local status state
* feat: Add logging and status sync for translation action
* feat: Refactor action component state management to be consistent with
translate action
Replace separate `isContented` and `isLoading` states with a single
`status` state that tracks 'preparing', 'streaming', and 'finished'
phases. Sync status with assistant message status and update footer
loading prop accordingly.
* fix: Add missing pauseTrace import to ActionTranslate component
* fix: Add missing break statements in assistant message status handling
* fix: Move pauseTrace call inside abort completion condition
* feat(mcp): add hub server type definitions
- Add 'hub' to BuiltinMCPServerNames enum as '@cherry/hub'
- Create GeneratedTool, SearchQuery, ExecInput, ExecOutput types
- Add ExecutionContext and ConsoleMethods interfaces
Amp-Thread-ID: https://ampcode.com/threads/T-019b4e7d-86a3-770d-82f8-9e646e7e597e
Co-authored-by: Amp <amp@ampcode.com>
* feat(mcp): implement hub server core components
- generator.ts: Convert MCP tools to JS functions with JSDoc
- tool-registry.ts: In-memory cache with 10-min TTL
- search.ts: Comma-separated keyword search with ranking
- runtime.ts: Code execution with parallel/settle/console helpers
Amp-Thread-ID: https://ampcode.com/threads/T-019b4e7d-86a3-770d-82f8-9e646e7e597e
Co-authored-by: Amp <amp@ampcode.com>
* feat(mcp): integrate hub server with MCP infrastructure
- Create HubServer class with search/exec tools
- Implement mcp-bridge for calling tools via MCPService
- Register hub server in factory with dependency injection
- Initialize hub dependencies in MCPService constructor
- Add hub server description label for i18n
Amp-Thread-ID: https://ampcode.com/threads/T-019b4e7d-86a3-770d-82f8-9e646e7e597e
Co-authored-by: Amp <amp@ampcode.com>
* test(mcp): add unit tests for hub server
- generator.test.ts: Test schema conversion and JSDoc generation
- search.test.ts: Test keyword matching, ranking, and limits
- runtime.test.ts: Test code execution, helpers, and error handling
Amp-Thread-ID: https://ampcode.com/threads/T-019b4e7d-86a3-770d-82f8-9e646e7e597e
Co-authored-by: Amp <amp@ampcode.com>
* docs(mcp): add hub server documentation
- Document search/exec tool usage and parameters
- Explain configuration and caching behavior
- Include architecture diagram and file structure
Amp-Thread-ID: https://ampcode.com/threads/T-019b4e7d-86a3-770d-82f8-9e646e7e597e
Co-authored-by: Amp <amp@ampcode.com>
* ♻️ refactor(hub): simplify dependency injection for HubServer
- Remove HubServerDependencies interface and setHubServerDependencies from factory
- Add initHubBridge() to mcp-bridge for direct initialization
- Make HubServer constructor parameterless (uses pre-initialized bridge)
- MCPService now calls initHubBridge() directly instead of factory setter
- Add integration tests for full search → exec flow
* 📝 docs(hub): add comments explaining why hub is not in builtin list
- Add JSDoc to HubServer class explaining its purpose and design
- Add comment to builtinMCPServers explaining hub exclusion
- Hub is a meta-server for LLM code mode, auto-enabled internally
* ✨ feat: add available tools section to HUB_MODE_SYSTEM_PROMPT
- Add shared utility for generating MCP tool function names (serverName_toolName format)
- Update hub server to use consistent function naming across search, exec and prompt
- Add fetchAllActiveServerTools to ApiService for renderer process
- Update parameterBuilder to include available tools in auto/hub mode prompt
- Use CacheService for 1-minute tools caching in hub server
- Remove ToolRegistry in favor of direct fetching with caching
- Update search ranking to include server name matching
- Fix tests to use new naming format
Amp-Thread-ID: https://ampcode.com/threads/T-019b6971-d5c9-7719-9245-a89390078647
Co-authored-by: Amp <amp@ampcode.com>
* ♻️ refactor: consolidate MCP tool name utilities into shared module
- Merge buildFunctionCallToolName from src/main/utils/mcp.ts into packages/shared/mcp.ts
- Create unified buildMcpToolName base function with options for prefix, delimiter, maxLength, existingNames
- Fix toCamelCase to normalize uppercase snake case (MY_SERVER → myServer)
- Fix maxLength + existingNames interaction to respect length limit when adding collision suffix
- Add comprehensive JSDoc documentation
- Update tests and hub.test.ts for new lowercase normalization behavior
* ✨ feat: isolate hub exec worker and filter disabled tools
* 🐛 fix: inline hub worker source
* 🐛 fix: sync hub tool cache and map
* Update import path for buildFunctionCallToolName in BaseService
* ✨ feat: refine hub mode system prompt
* 🐛 fix: propagate hub tool errors
* 📝 docs: clarify hub exec return
* ✨ feat(hub): improve prompts and tool descriptions for better LLM success rate
- Rewrite HUB_MODE_SYSTEM_PROMPT_BASE with Critical Rules section
- Add Common Mistakes to Avoid section with examples
- Update exec tool description with IMPORTANT return requirement
- Improve search tool description clarity
- Simplify generator output with return reminder in header
- Add per-field @param JSDoc with required/optional markers
Fixes issue where LLMs forgot to return values from exec code
* ♻️ refactor(hub): return empty string when no tools available
* ✨ feat(hub): add dedicated AUTO_MODE_SYSTEM_PROMPT for auto mode
- Create self-contained prompt teaching XML tool_use format
- Only shows search/exec tools (no generic examples)
- Add complete workflow example with common mistakes
- Update parameterBuilder to use getAutoModeSystemPrompt()
- User prompt comes first, then auto mode instructions
- Skip hub prompt when no tools available
* ♻️ refactor: move hub prompts to dedicated prompts-code-mode.ts
- Create src/renderer/src/config/prompts-code-mode.ts
- Move HUB_MODE_SYSTEM_PROMPT_BASE and AUTO_MODE_SYSTEM_PROMPT_BASE
- Move getHubModeSystemPrompt() and getAutoModeSystemPrompt()
- Extract shared buildToolsSection() helper
- Update parameterBuilder.ts import
* ♻️ refactor: add mcpMode support to promptToolUsePlugin
- Add mcpMode parameter to PromptToolUseConfig and defaultBuildSystemPrompt
- Pass mcpMode through middleware config to plugin builder
- Consolidate getAutoModeSystemPrompt into getHubModeSystemPrompt
- Update parameterBuilder to use getHubModeSystemPrompt
* ♻️ refactor: move getHubModeSystemPrompt to shared package
- Create @cherrystudio/shared workspace package with exports
- Move getHubModeSystemPrompt and ToolInfo to packages/shared/prompts
- Add @cherrystudio/shared dependency to @cherrystudio/ai-core
- Update promptToolUsePlugin to import from shared package
- Update renderer prompts-code-mode.ts to re-export from shared
- Add toolSetToToolInfoArray converter for type compatibility
* Revert "♻️ refactor: move getHubModeSystemPrompt to shared package"
This reverts commit 894b2fd487.
* Remove duplicate Tool Use Examples header from system prompt
* fix: add handleModeChange call in MCPToolsButton for manual mode activation
* style: update AssistantMCPSettings to use min-height instead of overflow for better layout control
* feat(i18n): add MCP server modes and truncate messages in multiple languages
- Introduced new "mode" options for MCP servers: auto, disabled, and manual with corresponding descriptions and labels.
- Added translations for "base64DataTruncated" and "truncated" messages across various language files.
- Enhanced user experience by providing clearer feedback on data truncation.
* Normalize tool names for search and exec in parser
* Clarify tool usage rules in code mode prompts and examples
* Clarify code execution instructions and update example usage
* refactor: simplify JSDoc description handling by removing unnecessary truncation
* refactor: optimize listAllActiveServerTools method to use Promise.allSettled for improved error handling and performance
---------
Co-authored-by: Amp <amp@ampcode.com>
Co-authored-by: kangfenmao <kangfenmao@qq.com>
* fix: resolve ActionTranslate stalling after initialization
Issue: When invoking translate from the selection assistant, the fetchResult function does not react to the completion of initialize, causing the ActionTranslate component to enter an infinite loading state.
Cause: In commit 680bda3993, the initialize effect hook was refactored into a callback function. This refactor omitted the notification that fetchResult should run after initialization, so fetchResult never executes post‑initialization.
Fix: Change the initialized flag from a ref to a state variable and have fetchResult listen to this state. This modification ensures the effect hook triggers fetchResult exactly once after initialization is complete.
* fix(ActionTranslate): fix missing dependency in useEffect hook
---------
Co-authored-by: icarus <eurfelux@gmail.com>
AI SDK v5 uses 'developer' role for reasoning models, but some providers
like Azure DeepSeek R1 only support 'system', 'user', 'assistant', 'tool'
roles, causing HTTP 422 errors.
This fix adds a custom fetch wrapper that converts 'developer' role back
to 'system' for providers that don't support it.
Fixes#12321🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude <noreply@anthropic.com>
Use fixed base URL for TokenFlux image API instead of provider.apiHost.
After migration 191, apiHost was changed to include /openai/v1 suffix
for chat API compatibility, but image API needs the base URL without
this suffix, causing /openai/v1/v1/images/models (wrong path).
Fixes#12284
Signed-off-by: SherlockShemol <shemol@163.com>
- Changed all 'Assistant:' labels to 'A:' in tool use examples for consistency
- Added missing blank line before final response in both files
- Affects promptToolUsePlugin.ts and prompt.ts
- Resolves#12310
* fix: use ipinfo lite API with token for IP country detection
Switch from ipinfo.io/json to api.ipinfo.io/lite/me endpoint with
authentication token to improve reliability and avoid rate limiting.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix: use country_code field from ipinfo lite API response
The lite API returns country_code instead of country field.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
