Pull Request Explorer

Summary

Refactor ToolTag to always use filled variant

Simplifies tag appearance by removing dark mode conditional, ensuring consistent UI across themes.

Health Assessment

• • Quick refactor with minimal changes, reviewed immediately, no rework needed.

Summary

Add scene-specific model config for follow-up extraction

Enables the onboarding agent to use its own model for generating follow-up suggestion chips, improving customization and reducing reliance on the generic system agent.

Health Assessment

• • Fast review with no rework, moderate code changes, low risk.

Summary

fix desktop token refresh retry

Prevents frequent logout from token refresh retry on desktop client

Health Assessment

• • The PR had a fast cycle time of 4.4 hours and received automated review suggestions from chatgpt-codex-connector

Summary

Restore route after desktop update restart

Ensures users return to their previous screen after the desktop app updates and restarts, improving continuity and user experience.

Health Assessment

• • Quick review and merge within 5.6 hours, minimal rework, indicating a straightforward bug fix.

Summary

Fix market 404 mapping to NOT_FOUND

Corrects error handling for first‑time market users, preventing noisy 500 logs and unnecessary retries. Ensures smoother onboarding and cleaner observability.

Health Assessment

• • Quick fix with minimal changes, resolved in under two hours with no rework.

Summary

Fix server‑side exec_task for async callAgent

Ensures sub‑agents are correctly spawned when a parent agent runs as a server‑side task, improving reliability of async agent calls. This change prevents silent failures and ensures proper task dispatching.

Health Assessment

• • The PR was reviewed quickly with minimal back‑and‑forth, indicating a straightforward bug fix that was well‑understood by the team.

Summary

docs: LobeHub Your Chief Agent Operator

Updated documentation for LobeHub's Chief Agent Operator, improving clarity for users.

Health Assessment

• • PR merged quickly with minimal review, indicating a straightforward documentation update.

Summary

Fix memory-agent message isolation

Ensures memory-agent messages are stored in a dedicated child thread, preventing main conversation timeline pollution and improving user clarity.

Health Assessment

• • Quick, single-commit fix with minimal code changes and no new tests, indicating low complexity and risk.

Summary

Fix schedule pill wrapping in Kanban view

Ensures schedule pills stay on a single line, preventing layout breakage in the Agent Tasks Kanban view and improving user experience.

Health Assessment

• • Quick fix with minimal changes, fast review and merge, low risk.

Summary

Fix sidebar new agent

The sidebar now correctly displays the new agent option, preventing confusion. This improves user experience when managing agents.

Health Assessment

• • The PR was reviewed and merged within minutes, indicating minimal friction and a straightforward fix.

Summary

Release v2.2.0

Introduces the Chief Agent Operator, a self-iterating agent that runs end-to-end, and various other features and fixes.

Health Assessment

• • The PR has a large scope with many changes, but the review process was relatively smooth and fast.

Summary

Fix memory receipt routing metadata persistence

Ensures memory receipt routing metadata is preserved after Redis persistence, improving accurate routing to memory detail pages.

Health Assessment

• • Quick turnaround with minimal rework; AI review bots provided suggestions but no major changes required.

Summary

Restore DeepSeek models to official pricing

Updates DeepSeek model pricing to official rates, ensuring accurate cost calculations for users.

Health Assessment

• • The PR was reviewed and merged quickly, indicating a smooth process.
• • The changes are limited to pricing data, reducing the risk of introducing new bugs.

Summary

Fix conversation animation and remove clear messages hotkey

This PR fixes the markdown streaming animation to only animate the last block and removes the Alt+Shift+Backspace hotkey for clearing messages, improving user experience and reducing unnecessary UI flicker.

Health Assessment

• • Quick fix with minimal changes, AI-assisted code generation, fast review turnaround.

Summary

Switch action tag chips to icon and label

Enhances chat input UI by replacing badge‑style tags with inline icons and colored labels, improving readability and user experience.

Health Assessment

• • PR completed in 1 hour with AI assistance and minimal review, indicating efficient process and low risk.

Summary

Fix Google functionCall thoughtSignature across turns

Ensures consistent thoughtSignature handling for cross-provider conversations, preventing API alerts.

Health Assessment

• • Quick fix with minimal changes, fast review, no rework.

Summary

Refactor chat input to use native submenu slots

Replaces sticky CSS hacks with native header/footer slots in the skill menu, improving UI stability and reducing duplicate elements.

Health Assessment

• • Quick merge with no review comments indicates a low-risk refactor.

Summary

Fix video image reference to JSON object

Ensures OpenAI-compatible video requests use proper JSON structure, improving API compatibility and preventing errors.

Health Assessment

• • Quick fix with minimal changes and immediate review, indicating low complexity and high confidence.

Summary

Add ernie-5.1 support

Adds support for the ernie-5.1 AI model to improve chat functionality.

Health Assessment

• • The PR was merged quickly, but the author initially hit a rate limit on their Sourcery AI account.

Summary

Fix auto‑collapsing right panel on chat mount

Prevents the right workspace panel from unexpectedly closing when switching topics, improving user experience.

Health Assessment

• • Quick review and merge within 7.4 hours, minimal rework.

Summary

Fix agent tool state restoration on multi-replica

Ensures correct parent-child relationships and model data for agents across replicas, preventing broken conversation flows.

Health Assessment

• • Fast turnaround with minimal review, indicating a straightforward bug fix; automated AI reviews provided suggestions but no major rework.

Summary

Add project skills UI and markdown preview

Enables users to view and interact with project skills in the sidebar and preview markdown files, improving workflow in heterogeneous Claude Code sessions.

Health Assessment

• • Rapid review and merge (under an hour) suggest high confidence, but the large number of lines and files indicates significant integration effort and potential for downstream impact.

Summary

Restore FullNameStep back button to shared prefix

Fixes broken back navigation in the onboarding flow, allowing users to return to previous steps correctly.

Health Assessment

• • Fast turnaround with minimal changes; no significant review friction or risk.

Summary

Fix agent-signal routing for hinted skill documents

Corrects routing logic so that hinted skill documents are treated as skill-domain evidence, improving agent skill management accuracy.

Health Assessment

• • Quick, single-commit fix with AI-assisted review; minimal code churn and fast approval.

Summary

Refactor agent signal prompts for language adaptation

Enhances prompt language handling and localization for agent signals, improving user experience across languages.

Health Assessment

• • Fast cycle time and minimal review iterations suggest high confidence in the refactor. The large code changes indicate significant scope but were well-received.

Summary

Add Chrome-style divider between inactive tabs

Improves tab visibility in the Electron desktop app by adding a subtle divider, reducing confusion when tabs blend into the titlebar.

Health Assessment

• • Fast cycle time of 18 minutes, minimal review comments, and AI-assisted code generation indicate a smooth, low-risk PR.

Summary

Fix local-system executor param forwarding and move executor to client

Ensures full grepContent parameters are forwarded to runtime, fixes silent parameter drops, and reorganizes executor code under client for consistency, improving reliability of local-system tool.

Health Assessment

• • Quick turnaround with minimal review, indicating straightforward bug fix with clear tests.

Summary

Fix per-tool timeout handling in agent runtime

Ensures long-running tool commands respect user-specified timeouts, preventing server-side failures and improving reliability for end users.

Health Assessment

• • The PR resolved a critical timeout issue with a single commit and rapid review, but the extensive code modifications raise implementation risk.

Summary

Share Linear inspector across CC and built‑in skill

Provides a consistent, branded UI for Linear tool calls from both the CC adapter and the built‑in skill, improving user experience and reducing confusion.

Health Assessment

• • Fast 1.6‑hour cycle with zero review rounds; AI‑assisted reviews helped surface issues early, leading to a smooth merge.

Summary

Improve agent signal prompts performance

Enhances prompt handling and rule clarity for agent signals, boosting performance and reliability.

Health Assessment

• • The PR was merged within 1.2 hours with minimal review, indicating a straightforward change.
• • The moderate code churn and single commit suggest low risk and quick delivery.

Summary

Translate non-English comments to English in agent-tasks

Non‑English comments in the AgentTasks feature were translated to English, preserving functionality and formatting.

Health Assessment

• • Fast review and merge with minimal changes, indicating low risk and high confidence in the update.

Summary

Add task tools and AskUserQuestion freeform note with waitingForHuman status

Introduces new task management tools and a user‑interaction feature that allows freeform responses, enhancing workflow control and human intervention handling in Lobe Chat.

Health Assessment

• • The PR was reviewed immediately and merged within 21 hours, indicating efficient collaboration and low friction despite its large scope.

Summary

Refactor agent settings: drop Meta/Documents tabs, restore inputTemplate

Simplifies agent settings UI by removing obsolete tabs and restoring the input template field, improving user experience and reducing confusion.

Health Assessment

• • Quick 3.6‑hour cycle with no review delays, minimal rework, and AI‑assisted code generation. No blockers or large scope.

Summary

Shorten usage token detail display

Improves readability of usage metrics by showing abbreviated values, enhancing user experience.

Health Assessment

• • Quick review and minimal changes indicate a straightforward style update with minimal risk.

Summary

Fix Gemini schema enum/required sanitization

Ensures Gemini tool schema complies with type constraints, preventing runtime errors and improving reliability.

Health Assessment

• • PR had rapid initial review and multiple commits to address feedback, indicating active iteration. Large code changes and extensive test additions suggest significant impact on Gemini integration.

Summary

Polish brief recommendations layout

Enhances user experience by hiding unnecessary headers, standardizing button layout, and adding tests to ensure stable recommendation display.

Health Assessment

• • The PR was merged within 24 minutes with no back‑and‑forth, indicating a straightforward change.
• • The addition of regression tests improves reliability without adding significant risk.

Summary

Support Monitor-style signal callbacks in AssistantGroup

Adds support for Monitor tool callbacks, grouping them into a single assistant message with collapsible UI, improving user experience by reducing clutter.

Health Assessment

• • The PR had a rapid review but required multiple iterations, indicating complex changes. The large scope and many commits after review suggest moderate risk.

Summary

Fix task schedule max executions and heartbeat limits

Ensures scheduled tasks respect max execution limits and enforces a minimum heartbeat interval, preventing runaway loops and improving reliability.

Health Assessment

• • The PR was reviewed and merged within 4 hours with minimal back‑and‑forth, indicating high confidence in the changes despite the large code churn.

Summary

Add Linear MCP tool inspector

Adds a branded inspector and summary for Linear MCP tool calls, improving user clarity and workflow visibility.

Health Assessment

• • Merged within 3.6 hours with no review rounds, indicating efficient review and high confidence in the AI-generated implementation.

Summary

Fix DB trace key mismatch for agent tracing

Aligns database trace keys with S3 object keys, preventing 404 errors in agent tracing UI.

Health Assessment

• • Quick fix with minimal changes, resolved in under 3 hours, no rework needed.

Summary

Health Assessment

Summary

Fix agent config fetch for UI toggles

Ensures that toggling files and knowledge bases in the home page immediately updates the UI, improving user experience and data consistency.

Health Assessment

• • Quick fix with minimal changes, no back‑and‑forth, resolved in under 18 hours.

Summary

Style subagent rows and remove badge

Improves readability of subagent thread items by replacing the tree icon and removing the redundant badge, ensuring a cleaner UI.

Health Assessment

• • Merged within 14 hours with no rework, indicating a straightforward style change.

Summary

Fix SchedulerForm race and stale refresh CLS

This fix prevents UI flicker and ensures accurate scheduling updates, improving user experience and reducing errors.

Health Assessment

• • Rapid resolution with minimal review indicates a straightforward bug fix and low complexity.

Summary

Fix editor placeholder behavior

Prevents misleading placeholder text from appearing after content is entered, improving user experience in task editing.

Health Assessment

• • Rapid resolution with single commit and immediate approval.

Summary

Fix enter-to-send preference in chat inputs

Aligns thread and task comment inputs with the global Cmd+Enter setting, ensuring consistent user experience across the chat interface.

Health Assessment

• • Fast turnaround with no rework; minimal code changes and no breaking impact.

Summary

Align Add Subtask button and card baseline

Improves visual alignment of the subtask button and card content, enhancing the user experience on the agent task detail page.

Health Assessment

• • Quick visual fix with minimal changes, AI-assisted, no rework needed.

Summary

Style chat input action bar padding

Adjust padding to equalize send button breathing room, improving UI consistency.

Health Assessment

• • Quick UI tweak with minimal changes, resolved in under 4 hours with no rework.

Summary

Enable send button after paste in comment input

Fixes UI bug where the send button remains disabled after pasting content without typing, improving user experience.

Health Assessment

• • The PR was reviewed and merged within 12 hours with minimal changes, indicating a straightforward fix with low risk.

Summary

Preserve TopicChatDrawer state during close animation

Fixes a UI flicker that occurs when closing the TopicChatDrawer, ensuring a smooth user experience by maintaining the drawer’s content until the animation completes.

Health Assessment

• • The PR was reviewed and merged within 24 hours with minimal changes, indicating a straightforward, low-risk bug fix.