feat(core): enable parallel tool call execution#2000
Draft
yiliang114 wants to merge 42 commits intosplit/tool-scheduler-refactorfrom
Draft
feat(core): enable parallel tool call execution#2000yiliang114 wants to merge 42 commits intosplit/tool-scheduler-refactorfrom
yiliang114 wants to merge 42 commits intosplit/tool-scheduler-refactorfrom
Conversation
Changed sub-agent tool calls from sequential to parallel execution using Promise.allSettled() to significantly improve performance when the model returns multiple tool calls in a single response. Key changes: - Use Promise.allSettled() for parallel tool call execution - Extract executeSingleToolCall() method for better code organization - Maintain existing state management and error handling - Add parallel execution timing tests Parallelism is safe because LLM tool call protocol guarantees that tool calls in the same response cannot reference each other's outputs.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
…d prioritize project-level settings - SubAgentScope now automatically appends global memory (including QWEN.md and output-language.md) to its system prompt, ensuring subagents respect project conventions and language preferences. - The CLI config loader now checks for `.qwen/output-language.md` in the project directory before falling back to the global `~/.qwen/output-language.md`, allowing project-specific language settings. Made-with: Cursor
…dealtalk-api-inner
…1305 - Add error code 1305 to RATE_LIMIT_ERROR_CODES for DashScope/IdealTalk internal rate limit detection (issue #1918) - Add test case for 1305 error code detection - Update existing test cases to use 9999 as custom error code to avoid conflict - Remove unused sdk-java/.gitignore file Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
…pt for project conventions)
…scheduler-parallel
Contributor
📋 Review SummaryThis PR enables parallel execution of independent tool calls within the same batch using 🔍 General Feedback
🎯 Specific Feedback🟡 High
🟢 Medium
🔵 Low
✅ Highlights
|
…guage path priority - Add 3 tests for SubAgentScope.buildChatSystemPrompt appending userMemory - Add 3 tests for project-level output-language.md path resolution priority Made-with: Cursor
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Add AuthDisplayType enum and helper for Coding Plan detection - Remove formatAuthType/titleizeAuthType functions - Update tests for new auth types Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Add i18n keys for modality types and status labels - Update ModelDialog to use t() for user-facing strings Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Update error messages for unsupported image/PDF inputs with clearer guidance - Add `modalities` setting to override auto-detected input modalities - Document `modalities` config in model-providers.md and settings.md - Update converter tests to match new error message format This provides users with actionable alternatives when their selected model doesn't support certain input types, and allows manual modality overrides for models not recognized by auto-detection. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com> - Add 5MB limit for image files to prevent API errors - Add 10MB limit for PDF files based on provider constraints - Return FILE_TOO_LARGE error with clear message when limits exceeded - Add tests for both image and PDF size limit enforcement This prevents errors when attempting to process large binary files that exceed provider API limits.
Add AGENT_CONTEXT_FILENAME constant and include AGENTS.md in default context file search alongside QWEN.md. This enables out-of-the-box support for the AGENTS.md standard (https://agents.md/) without requiring manual configuration. Changes: - Add AGENT_CONTEXT_FILENAME = 'AGENTS.md' constant - Update currentGeminiMdFilename default to include both AGENTS.md and QWEN.md - AGENTS.md takes precedence (first in array) for the de facto standard Closes #2006
- Add weighted tip system to make certain tips appear more frequently - Set /insight tip to weight 3 (3x more likely than regular tips) - Add i18n translations for the new tip across all supported languages - Add comprehensive unit tests for weighted tip selection Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
feat(cli): Increase /insight feature exposure via weighted tips
- Removed exact match assertion for help text that changed in UI - Test now only verifies the dialog title renders correctly The help text changed from 'Enter to select · Esc to close' to 'Enter to select, ↑↓ to navigate, Esc to close', causing the test to fail unnecessarily. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
…heck - Reduce general file size limit from 20MB to 10MB (using 9.9MB threshold) - Remove per-type size limits (5MB images, 10MB PDFs) - Add base64 encoding size check for PDFs to prevent data URI limit errors - Update all tests to reflect new 10MB limit This fixes issue #1880 where large PDFs could exceed API data URI limits after base64 encoding, causing errors. The 9.9MB threshold provides margin for encoding overhead. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com> Replace generic skill suggestion with specific guidance to install the document-skills extension for PDF processing.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
chore: bump version to 0.11.1
Keep QWEN.md as the first element in the default context filenames array to maintain backward compatibility with /init command and save_memory tool. Both QWEN.md and AGENTS.md will still be searched and loaded, but QWEN.md takes precedence for file creation operations.
…logs Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
feat: support AGENTS.md as default context file
fix(subagent): append output-language.md to subagent system prompt and prioritize project-level settings
Add missing 'modes' and 'configOptions' fields to NewSessionResponse to enable mode switching (plan/yolo/auto-edit) in Zed ACP client. - Update schema.ts: add configOptionSchema and extend newSessionResponseSchema - Update acpAgent.ts: add buildModesData() and buildConfigOptions() helpers Fixes ACP mode control issue where Zed couldn't switch approval modes. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Update session/list response to match Claude Agent's format: - Change 'items' to 'sessions' array - Add 'title' field (mapped from prompt) - Add 'updatedAt' field (ISO format from mtime) - Remove 'hasMore' and 'nextCursor' fields This fixes session list display in Zed ACP client. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Add sessionCapabilities with fork, list, and resume capabilities to initialize response to match Claude Agent's format. This enables Zed ACP client to discover session management features. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Make cwd optional in listSessionsRequestSchema (defaults to process.cwd()) - Remove fork from sessionCapabilities (not supported) Fixes -32602 error when Zed calls session/list without cwd param. Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
fix(core/rateLimit): add support for rate limit error code 1305 and custom retry error codes
fix(logging): reduce excessive streaming output in session history logs
docs: update installation instructions
fix: add modality defaults to prevent API errors when reading PDFs and other media
Fix ACP protocol compatibility issues with Zed editor
* feat: add Ctrl+Y shortcut to retry failed requests
- Add Ctrl+Y keyboard shortcut for retrying the last failed request
- Add isNetworkError() to detect transient network failures (ECONNREFUSED, ETIMEDOUT, etc.)
- Add DashScope 1305 error code to rate limit detection
- Add error hint \"Press Ctrl+Y to retry\" in error messages
- Support user-defined error codes for retry via config
- Add retryLastPrompt() hook in useGeminiStream
- Update keyboard shortcuts documentation
* feat: improve Ctrl+Y retry feature with tests, docs, and rate limit config
- Add comprehensive tests for Ctrl+Y retry shortcut in InputPrompt
- Add unit tests for retryLastPrompt in useGeminiStream hook
- Add detailed JSDoc comments for retryLastPrompt function and Ctrl+Y shortcut
- Extend isRateLimitError to support custom error codes via retryErrorCodes config
- Fix rate limit retry log variable reference (RATE_LIMIT_RETRY_OPTIONS → maxRateLimitRetries)
- Add Eclipse IDE files to .gitignore
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* refactor(ui): consolidate retry countdown as inline hint in error messages
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* feat(cli): enhance error handling with improved retry mechanism
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Modify ErrorMessage component to remove dim color from hint text
- Update useGeminiStream hook to improve retry countdown behavior with option to preserve or clear hints
- Adjust tests to match new error handling implementation
* feat: add Ctrl+Y shortcut to retry the last failed request
When a request errors out, the error message shows an inline hint
"(Press Ctrl+Y to retry.)" in secondary color. Pressing Ctrl+Y
re-submits the same prompt, commits the error text to history
(without the hint), and clears the hint from the UI.
- Add retryLastPrompt action wired to Ctrl+Y via keyBindings and InputPrompt
- Track last submitted prompt and error state in useGeminiStream refs
- Show retry hint inline with error text in ErrorMessage component,
wrapping naturally on narrow terminals while preserving hint color
- Expose retryLastPrompt through UIActionsContext
- Add keyboard shortcut entry in KeyboardShortcuts display
- Add i18n strings for hint and no-retry-available message
- Document Ctrl+Y in keyboard-shortcuts.md
* docs(configuration): Update model provider configuration document
* chore: remove YOLO mode code from core
* fix: prevent Ctrl+Y hint from overriding auto-retry countdown
When an auto-retry countdown is active (retryCountdownTimerRef is set),
handleErrorEvent should not overwrite it with the Ctrl+Y hint. The auto-retry
hint ("retrying in Xs...") and manual retry hint ("Press Ctrl+Y to retry.")
are mutually exclusive:
- Auto-retry errors (e.g., rate limits): show countdown hint
- Other errors: show Ctrl+Y hint
Also removed retryErrorCodes from ContentGeneratorConfig as it's not part
of the minimal Ctrl+Y feature scope.
* simplify: remove complex options from clearRetryCountdown
Revert clearRetryCountdown to simplest form without options parameter.
The function now just clears the timer and pending item without any
automatic history commit logic.
* fix: restore pendingRetryCountdownItem as separate state from pendingRetryErrorItem
Auto-retry countdown and manual retry hint are now independent:
- pendingRetryErrorItem: displays error message with optional hint
- pendingRetryCountdownItem: displays separate countdown line for auto-retry
This ensures both can be shown simultaneously without overriding each other.
* fix: restore RetryCountdownMessage rendering in HistoryItemDisplay
The retry_countdown type should be rendered as a separate message,
not inline in ErrorMessage. This allows auto-retry countdown and
manual retry hint to coexist properly.
* fix(cli): properly commit retry error item to history before clearing
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* fix(cli): remove trailing period from retry hint translations
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Remove unnecessary period from 'Press Ctrl+Y to retry' translation strings in both en.js and zh.js locales. Also update the corresponding usage in useGeminiStream hook.
* chore(sdk-java): add Eclipse project configuration files
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Add .project configuration files for client and qwencode modules to support Eclipse IDE development environment.
* feat(cli): add retry countdown hint to error message
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* Revert "chore(sdk-java): add Eclipse project configuration files"
This reverts commit da83b5e.
---------
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
refactor(core): extract single tool-call execution path
…/tool-scheduler-parallel
…ential execution guidelines Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
This was referenced Mar 3, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Closes #1998
This PR enables parallel execution of independent tool calls within the same batch. Built on top of #1998 (refactor), it switches from sequential to concurrent execution using
Promise.allSettled.Changes
for...ofloop withPromise.allSettledfor parallel executionDepends On
Stack