Pull Request Explorer

Summary

Fix markdown footnotes rendering without citations

Ensures footnotes in assistant messages render correctly when no structured citations are present, improving clarity of responses.

Health Assessment

• • Quick, single‑commit fix with no rework, minimal risk to production.

Summary

Make linked PR clickable in meta hover card

Adds clickability to the PR link in the topic hover card, enabling users to open the PR directly from the UI.

Health Assessment

• • Quick review and merge with minimal changes; no regressions reported.

Summary

Trim homepage batch payload for performance

Reduces initial TRPC batch size by trimming over-fetches, improving load times for users.

Health Assessment

• • Fast review and merge within 1.2 hours indicates a smooth process with minimal friction.

Summary

Fix todo defaultItems guard against non-array args

Prevents crashes caused by malformed tool‑call arguments in the lobe‑agent todo list, improving stability for users.

Health Assessment

• • Quick, single‑commit fix with AI assistance; minimal review and no rework required.

Summary

Fix topic spinner stuck after local CC run finishes

Corrects the UI spinner that remains active after a local Claude Code run completes, ensuring the spinner clears promptly and improving user experience.

Health Assessment

• • Quick fix with minimal changes, resolved in a single commit after immediate review, indicating low complexity and high confidence in the patch.

Summary

Fix Agent Builder identity field guard

Prevents accidental overwriting of the Agent Builder’s own data by stripping protected fields during updates, ensuring data integrity and preventing dirty rows.

Health Assessment

• • The PR was merged within 30 minutes of opening, with a single AI‑reviewed commit and no back‑and‑forth, indicating a straightforward, low‑risk fix.

Summary

Fix device-scope skill preview on desktop

Enables preview of device-scoped skills in the desktop UI, improving developer workflow.

Health Assessment

• • Quick fix with minimal changes, no rework, AI-assisted code generation.

Summary

Expose Composio router via mobile tRPC

Allows mobile clients to call composio procedures via /trpc/mobile

Health Assessment

• • Fast review and merge, minimal changes, low risk

Summary

Add mock workflow start helper to tests

Adds missing ensureWorkflowStarted export to mock runGuard, enabling fully mocked tests for processUserTopicsHandler.

Health Assessment

• • Merged within 15 minutes after a single review comment, minimal code changes, no blockers.

Summary

Fix workflow start order before guard checks

Ensures memory workflows persist an initial step, preventing retry loops and improving reliability for users.

Health Assessment

• • The PR was reviewed and merged within 36 minutes, with minimal rework and no major blockers. AI-assisted review helped catch edge cases early.

Summary

Cap per-user topic fan‑out and serve‑side flow control

This PR limits per‑user fan‑out to prevent message storms, replaces context.invoke with fire‑and‑forget triggers, and adds serve‑side flow control to improve reliability and resource usage.

Health Assessment

• • Quick AI‑assisted fix resolved unbounded fan‑out issue with minimal rework and fast review cycle.

Summary

Fix desktop login blank screen

Prevents a brief full‑screen blank flash during desktop login by making CacheHydrationGate a one‑way latch, improving user experience.

Health Assessment

• • Quick fix with minimal rework, fast review, low risk.

Summary

Fix spine promotion for cold replicas

Ensures agent state consistency across serverless replicas, preventing missed promotions and potential data loss.

Health Assessment

• • Quick, single‑commit fix with no rework; fast review and merge cycle indicates low complexity and minimal risk.

Summary

Fix memory guard Redis checks in workflows

Ensures memory workflows can be stopped at each step, improving reliability and operator control.

Health Assessment

• • Quick turnaround with minimal review, indicating a straightforward bug fix that required only a single review cycle.

Summary

Add askUserQuestion and disable agent onboarding

Enables agents to ask clarifying questions while removing the onboarding flow, improving user experience and reducing friction.

Health Assessment

• • The PR was merged within 6 hours with only a single review comment, indicating rapid developer confidence, but the large code changes and multiple commits suggest a moderate risk of integration complexity.

Summary

Fix hetero signal turn spine promotion

Corrects a rendering bug where tool‑bearing signal turns were not advancing the conversation spine, ensuring accurate trace and improved user experience.

Health Assessment

• • Fast turnaround with immediate review and a single commit, indicating a straightforward bug fix with minimal risk.

Summary

Fix multi-agent mention delegation in gateway mode

Ensures that when multiple agents are mentioned in gateway mode, the supervisor correctly forwards the mention data to the server, enabling each agent to run and respond, improving user experience and preventing silent failures.

Health Assessment

• • Fast turnaround with minimal rework; AI-assisted code generation likely contributed to quick resolution.

Summary

Persist working directory metadata

Adds persistent metadata for worktree context, improving session continuity and UI insights for users.

Health Assessment

• • PR required multiple commits to address AI tool limits and edge cases, indicating moderate complexity and high scope. The review was immediate but the code changes were extensive, reflecting a significant feature addition.

Summary

Add UI refinements to verify report page

Improves user experience by making the report pane white, enabling inline video playback, and categorizing evidence markers, enhancing clarity and usability.

Health Assessment

• • Fast turnaround with AI-generated code and automated reviews, minimal human intervention.

Summary

Add run guard to persona update handler

Adds a run guard to the persona update-writing handler, preventing unguarded execution during memory storms and ensuring graceful disabling via user or path-scoped guards.

Health Assessment

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

Summary

Fix empty state flash during task list refetch

Prevents a brief flicker of an empty state, improving user experience by showing a loading skeleton instead of an empty placeholder during task list refetches.

Health Assessment

• • The PR was reviewed and merged within 7.6 hours with minimal changes, indicating a straightforward bug fix with low risk and high confidence in stability.

Summary

Fix run guard to prevent retry storm

Stops infinite retry storms caused by guard errors, improving reliability and reducing costs.

Health Assessment

• • Fast turnaround with minimal rework; AI-assisted review helped identify and resolve the issue quickly.

Summary

Polish run card with rich text, colors, per-run follow-up

Enhances the run card UI to display rich‑text results, color‑coded content, and inline follow‑up replies, improving user experience and interaction.

Health Assessment

• • Fast turnaround with minimal review, indicating a straightforward UI polish that required little rework.

Summary

Reduce Redis reads in workflow guard

Optimizes workflow guard by consolidating Redis calls, improving performance and reducing latency.

Health Assessment

• • Merged within minutes after a single AI‑reviewed comment, the PR made a small, low‑risk performance improvement.

Summary

Fix quota cache preservation by source

Ensures users see accurate quota information and friendly messages when rate limits occur, improving reliability and user experience.

Health Assessment

• • Quick turnaround with AI-assisted review and automated tests, minimal friction.

Summary

Revert file proxy to restore public access

Restores public access to file URLs, ensuring shared links remain functional for users without requiring authentication.

Health Assessment

• • Quick revert resolved a regression with minimal review, indicating efficient collaboration.

Summary

Add subscription payment failed notification locales

Adds localized copy for subscription payment failure notifications, enabling users to receive alerts in their language. This improves user experience during payment issues.

Health Assessment

Summary

Fix device-only builtin skill gating and add scraping guidance

Ensures device-capable runs correctly expose the agent-browser skill and improves scraping reliability by adding dynamic page and anti-bot escalation guidance, enhancing user experience and system robustness.

Health Assessment

• • The PR underwent rapid iteration with multiple automated reviews and seven commits, indicating complex changes and high review friction.

Summary

Add workspace private resource transfer and copy

Enables teams to transfer and copy private workspace resources, improving data management and collaboration.

Health Assessment

• • Fast cycle and review, but the large scope indicates a significant change; AI review helped identify issues early.

Summary

Add Case 8 to agent-testing skill common-mistakes

Adds a new case to the agent-testing skill documentation, ensuring screenshot requests default to a full isolated environment run, improving reliability and user experience.

Health Assessment

• • Rapid review and merge with minimal changes, indicating a straightforward documentation update.

Summary

Make worktree removal non-blocking

Improves user experience by ensuring the confirm dialog closes immediately when removing a git worktree, running the operation asynchronously and providing toast notifications for success or failure.

Health Assessment

• • Quick turnaround with no rework, minimal review friction, and a small scope indicates a low-risk, high-impact UI improvement.

Summary

Fix workflow run guard cancellation via QStash REST

Ensures correct payload format for QStash API, preventing cancellation failures and improving reliability of workflow run guard cancellations.

Health Assessment

• • Quick fix with minimal iterations, AI-assisted review, merged within 15 minutes.

Summary

Add Upstash workflow metrics

Adds OpenTelemetry metrics for Upstash Workflow and QStash events, enabling better observability of workflow execution.

Health Assessment

• • Rapid iteration with automated review, but large code changes across backend and observability modules.

Summary

Expose run guard webhook route

Adds a webhook route for workflow run guards, enabling deployments to set run guards via the generic workflow capability.

Health Assessment

• • Merged within 0.8 hours with no review comments, indicating a straightforward refactor with minimal risk.

Summary

Show run last message on run card

Adds the ability to display the last assistant message on the task detail run card, giving users immediate context of the conversation outcome.

Health Assessment

• • The PR was reviewed and merged within 1.5 hours with a single commit, indicating a smooth process and minimal friction. The use of AI assistance likely contributed to the rapid turnaround.

Summary

Fix single @agent mention routing in gateway mode

Ensures that single agent mentions in gateway mode correctly trigger the target agent, improving chat flow and user experience.

Health Assessment

• • Quick fix with minimal changes, no server side impact, low risk

Summary

Make verify report TLDR full-width

Adjusts the verify report summary to span full width and use primary text color, improving readability.

Health Assessment

• • AI-generated style change merged within minutes, minimal review needed.

Summary

Refactor task detail to exclude brief activities

This change removes brief-type activities from task detail responses, simplifying the activity feed and reducing data payload for clients.

Health Assessment

• • The PR was merged quickly with minimal review, and AI-assisted code generation helped streamline the implementation.

Summary

Fix workflow guard cancellation by URL prefix

Enables broader workflow run cancellation using URL prefix, removing the need for log scans and improving cancellation reliability.

Health Assessment

• • Quick, single‑commit fix with AI‑assisted review and no rework; minimal impact on codebase and low risk.

Summary

Configure memory topic parallelism

Improves performance by configuring memory user-topic parallelism in workflows

Health Assessment

• • Fast review and merge, low number of files changed and lines added/deleted

Summary

Refine ActionBar working-directory UX

Improves user experience for the working-directory picker and git switchers in chat input by adding default selection, search, undo, and better error handling.

Health Assessment

• • Fast cycle time and immediate review indicate an efficient process with minimal rework; the changes are focused on UX improvements and do not introduce significant risk.

Summary

Fix working-sidebar drag-resize functionality

Restores ability to resize the right panel in the agent sidebar, improving user control and interface flexibility.

Health Assessment

• • Quick fix with minimal changes, resolved in under 20 minutes, indicating low complexity and high confidence.

Summary

Polish verification report list and evidence rendering

Enhances the verification page UI by improving number formatting, removing unnecessary file names, fixing layout issues, adding loading skeletons, and preserving image aspect ratios, resulting in a cleaner, more user-friendly experience.

Health Assessment

• • The PR was merged within 3.2 hours with no review comments, indicating a smooth process. The AI-generated code was accepted without issues, and the changes are purely UI refinements.

Summary

Fix Home recents isolation by workspace

Ensures recent items are correctly scoped to the current user and workspace, preventing data leakage across identity contexts.

Health Assessment

• • The PR was reviewed immediately but required a large rework after 5 days, indicating a delayed fix but minimal review friction.

Summary

Add device-scoped filesystem skills

Enables users to manage and activate skills per device, improving workflow flexibility and reducing conflicts between project and device skill sets.

Health Assessment

• • PR merged quickly with minimal review, indicating straightforward implementation and strong test coverage.

Summary

Add errored state for verifier infra failures

Allows the system to distinguish verifier crashes from actual delivery failures, preventing false repair loops and improving task accuracy.

Health Assessment

• • Fast cycle time and minimal review iterations indicate smooth integration; the change adds new enum and logic but is well-tested.

Summary

Add Codex-style two-pane diff + file-tree layout to review panel

Introduces a two-pane review panel with a diff list on the left and a directory tree on the right, enhancing navigation and user experience for reviewing changes.

Health Assessment

• • Fast turnaround with a single commit and immediate approval, indicating a straightforward implementation with minimal review friction.

Summary

Fix tasks-list composer, scheduled group, row noise

Improves task list UX by fixing composer behavior, grouping scheduled tasks, and removing redundant row info, enhancing clarity and reducing user confusion.

Health Assessment

• • Rapid review and merge within 1.1 hours, minimal rework, indicating a smooth process.

Summary

Docs: land tasks-list L2 audit findings into ux checklists

Adds new UX audit rules and updates documentation to reflect visual audit findings, improving user experience consistency.

Health Assessment

• • Fast merge with minimal review, indicating straightforward documentation update.

Summary

Fix folded turn final answer visibility

Ensures that finished chat turns always display their final answer outside the collapsed process, improving user experience when reviewing conversation history.

Health Assessment

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