Pull Request Explorer

Summary

Docs: Skill metadata cleanup and argument hints

Standardizes skill configuration metadata, improving auto‑trigger accuracy and user clarity without affecting runtime behavior.

Health Assessment

• • Rapid review and merge with minimal comments indicate straightforward documentation changes; large diff size but no code changes reduces risk.

Summary

Fix double updateHotkey on clear

Prevents duplicate API calls and toast messages when clearing hotkey input, improving user experience.

Health Assessment

• • Quick fix with minimal changes, reviewed and merged within hours, indicating low complexity and high confidence.

Summary

Refactor onboarding to merge agent marketplace

Consolidates agent marketplace APIs into the web onboarding tool, simplifying tool identifiers and reducing duplication.

Health Assessment

• • Fast review and merge with no back‑and‑forth; high code churn but straightforward refactor with no external breaking changes.

Summary

Fix inactive OIDC access rejection

Ensures banned or inactive OIDC users cannot obtain access, improving security and compliance.

Health Assessment

• • Rapid resolution with minimal review, indicating a straightforward bug fix.

Summary

Add Render for onboarding tools

Introduces UI components that replace raw debug tables with rich, user‑friendly cards for onboarding tool responses, enhancing clarity and engagement.

Health Assessment

• • The PR was reviewed and merged within 3 hours, with minimal comments and no rework, indicating a smooth, low‑friction process.

Summary

Refactor Knowledge Base RAG Runtime Sharing

Consolidated knowledge‑base RAG logic into a shared service, eliminating duplicate code and simplifying maintenance across client and server. This reduces code churn and potential bugs while ensuring consistent behavior for knowledge‑base searches.

Health Assessment

• • Fast review and merge with minimal iterations; code duplication removed, improving maintainability and reducing future technical debt.

Summary

Refactor bot to close activator bypass

Improves device access security by centralizing device tool registry and tightening checks, preventing unauthorized external activations.

Health Assessment

• • Fast cycle time and immediate review indicate high priority security fix; multiple commits suggest iterative refinement; overall risk moderate due to large scope but quick resolution.

Summary

Refactor task config to snapshot agent model

Pin agent's model at task creation to prevent future changes from affecting existing tasks, ensuring stable execution. This improves reliability and predictability of task runs.

Health Assessment

• • Fast cycle time with minimal human review, but large code changes and AI-generated code may increase risk of subtle bugs.

Summary

chore: drop task template tracking

Remove analytics tracking for task templates in preparation for UI redesign, cleaning up code and tests.

Health Assessment

• • Fast cycle time and minimal review indicate a straightforward cleanup; AI review comments helped surface potential issues early.

Summary

Drop unreachable aihubmix empty-apiKey test

Remove test that asserts impossible contract, simplifying test suite and preventing false failures.

Health Assessment

• • Quick merge with no rework, minimal scope, low risk.

Summary

Add reasoning_effort support for Grok 4.3

Enhance model capabilities with reasoning effort settings for improved performance

Health Assessment

• • Quick review and merge process
• • Moderate code changes with multiple file updates

Summary

Increase chat topic title length

Extends visible chat topic titles to improve identification and reduce truncation.

Health Assessment

• • Quick review and merge with minimal changes, indicating low complexity and risk.

Summary

Use full models endpoint for complete catalog

Ensures LobeHub instances retrieve the full AiHubMix model catalog, improving model availability and reliability.

Health Assessment

• • Fast review but long cycle due to initial delay; minimal rework; straightforward bug fix.

Summary

Add required activation reason for activator

Enforces that activation calls include a reason, improving transparency and user control.

Health Assessment

• • Fast review with no rework; single commit and minimal changes indicate a straightforward, low‑risk enhancement.

Summary

Fix onboarding early exit marketplace skip

Improves user experience by skipping marketplace handoff on early exit and removing hardcoded CJK prompts, ensuring smoother onboarding flow.

Health Assessment

• • The PR was reviewed and merged within 40 hours with minimal changes, indicating a straightforward, low-risk fix.

Summary

Consolidate self-review logic in agent signal

This PR consolidates self-review logic, improving maintainability and reducing duplication.

Health Assessment

• • Fast cycle time and minimal review iterations indicate efficient implementation.

Summary

Add read‑only SubAgent thread UI with breadcrumb

Provides a clearer, read‑only view for SubAgent threads, adding breadcrumb navigation and a thread switcher to improve user flow and reduce confusion.

Health Assessment

• • Fast turnaround with minimal review friction; merged within 1.2 hours after a single review cycle.

Summary

Show skeleton in action bar during config load

Adds a loading skeleton to the chat input action bar to prevent premature actions while configuration loads, improving user experience.

Health Assessment

• • The PR was merged within 0.9 hours with no review rework, indicating a smooth, low-risk change. AI assistance was used to author the code, but the impact is a minor UI improvement.

Summary

Add Recommendations module with hetero agent library

Introduces a new Recommendations section on the home page that surfaces actionable heterogenous agent options, enhancing user onboarding and agent adoption.

Health Assessment

• • Merged within 11.6 hours with no review rework, indicating a smooth, low-risk feature addition.

Summary

Optimize system bot

Adds and updates localization files for multiple languages, improving international user experience.

Health Assessment

• • Fast cycle time and immediate review indicate an efficient process; AI review bots provided automated feedback but did not block the merge.

Summary

Add AskUserQuestion tools for claude code

Foundation layer for interactive AskUserQuestion in the CC integration, introducing local MCP server and per-op bridge

Health Assessment

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

Summary

Improve agent signal performance and structure

This PR optimizes agent signal handling, tightening structure and fixing minor self‑review bugs, leading to faster, more reliable evaluations.

Health Assessment

• • Fast review and merge with minimal back‑and‑forth, but large code and test changes indicate significant effort.

Summary

Wrap long tool-call params instead of truncating

Tool‑call parameter values are now fully displayed instead of truncated, improving readability. This enhances user experience when reviewing tool call details.

Health Assessment

• • Fast 3‑hour cycle time with minimal review friction and a small scope of 65 lines indicates a low‑risk, quick visual fix.

Summary

Format tool execution time display

Improves readability of tool execution times by formatting minutes and seconds consistently with app style.

Health Assessment

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

Summary

Enrich stream parse errors with provider context

Adds provider/model context and parse position to stream errors, improving diagnostics for OpenAI/Anthropic SDK failures.

Health Assessment

• • Fast cycle time and minimal review rounds indicate a straightforward fix; AI-assisted code generation likely contributed to the quick turnaround.

Summary

Fix markdown links in daily-brief input placeholder

Removes markdown link syntax from the daily-brief hint placeholder while preserving link data for AI processing, improving user experience and preventing display issues.

Health Assessment

• • Quick turnaround with minimal rework; single review and merge within 2 hours indicates smooth process.

Summary

Fix server runtime for visual content parts

Consume text content parts in server-side Lobe Agent visual analysis runtime, preserving existing onText path.

Health Assessment

• • Quick review and merge with automated review from Sourcery AI.

Summary

Gate device tools by sender identity

Adds security gating for device tools, ensuring only authorized bot owners or first‑party users can invoke local or remote device operations, preventing accidental data leaks and unauthorized access.

Health Assessment

• • Rapid 1.5‑hour cycle time and a single review iteration indicate high confidence in the changes; the PR addresses a critical security flaw with comprehensive tests and audit logging.

Summary

Fix onboarding interest storage

Store predefined onboarding interests as canonical keys instead of localized labels, preserving custom interest text and task-template recommendation compatibility.

Health Assessment

• • The PR has a large scope with 20 files changed, but the review process was relatively smooth with a fast cycle time of 5.9 hours.

Summary

Remove WeChat URL rules from web crawler

Eliminates obsolete WeChat URL handling, simplifying crawler logic and reducing maintenance.

Health Assessment

• • Fast review and merge with minimal changes; AI-assisted code removal streamlined the process.

Summary

Translate non-English strings to English in scripts

Ensures consistent English messages across CLI, device gateway, and desktop scripts, improving user experience for English-speaking users.

Health Assessment

• • Quick turnaround with no review comments, indicating a straightforward change.

Summary

Fix hetero-agent sync across replicas

Ensures new-step assistant state is consistently replicated across agent instances, improving reliability.

Health Assessment

• • Rapid fix with minimal rework; AI review provided suggestions but no major changes required.

Summary

Remove obsolete cron job from lobehub

Eliminates legacy cron job, reducing maintenance overhead and potential scheduling conflicts. Ensures cleaner scheduling logic for future cron tasks.

Health Assessment

• • Fast cycle time and minimal review iterations indicate efficient resolution of a small bug fix, despite large line deletions.

Summary

Fix DB refresh on every ingest call

Ensures consistent content state across serverless instances, preventing data truncation and improving reliability for concurrent batch processing.

Health Assessment

• • Quick fix with minimal rework; AI review helped identify concurrency issue, leading to fast resolution and low risk.

Summary

Fix auto-decline of AskUserQuestion in Claude Code

This change stops the CLI from auto-rejecting AskUserQuestion calls, saving token usage and avoiding broken tool cards. It improves user experience by ensuring questions are presented correctly.

Health Assessment

• • Fast review and merge with minimal changes; no rework needed.

Summary

Guard readFile against binary blobs and oversized output

Adds safeguards to prevent large or binary files from being read, protecting LLM calls and system stability.

Health Assessment

• • Fast 11‑hour cycle with minimal review friction; multiple commits addressed new guard logic and test refactoring, indicating efficient collaboration.

Summary

Add user activity business hook

Adds a generic hook that triggers after user activity updates, preserving timestamps for downstream processing, improving business logic consistency.

Health Assessment

• • Fast AI‑driven review and minimal human intervention led to a quick 10‑hour cycle with only three commits after the initial review.

Summary

Refactor SiliconCloud model sync and deduplication

Synchronizes the SiliconCloud model list with the API, removes duplicate entries, and updates reasoning parameters to improve model accuracy and consistency.

Health Assessment

• • The PR required multiple merge commits to integrate with the canary branch, indicating integration challenges; despite a fast initial review, the large diff and repeated merges suggest significant integration friction.

Summary

i18n: update banner copy translations

Adds banner copy translations for multiple languages, enhancing international user experience.

Health Assessment

• • Merged instantly with no review, minimal changes, low risk.

Summary

Remove trailing punctuation from banner titles

Improves i18n by removing trailing punctuation from banner titles for better user experience.

Health Assessment

• • This PR was merged immediately, suggesting a straightforward and low-risk change.

Summary

Add Gemini 3.1 Flash-Lite provider cards

Adds support for the new Gemini 3.1 Flash-Lite model, enabling users to access improved multimodal capabilities and updated pricing.

Health Assessment

• • Fast cycle time of 9.2 hours and no review approvals suggest low complexity and minimal friction.

Summary

Refactor: Remove model extend param options

Simplifies model configuration by removing obsolete extend options, improving maintainability. Maintains backward compatibility with legacy configs and supports DeepSeek V4 reasoning settings.

Health Assessment

• • The PR was reviewed and merged within 8.4 hours, indicating a smooth process with minimal friction.
• • The change is a straightforward refactor with moderate code churn but no critical risk.

Summary

Add intro and screenshot to task scheduler changelog

Enhances the task scheduler changelog entry with a concise introduction and illustrative screenshot, improving user understanding of the new desktop‑only feature.

Health Assessment

• • Quick review and merge with minimal changes indicates a straightforward documentation update.

Summary

Fix user memory query normalization

Corrects user memory queries to use Paradedb match, improving accuracy and maintainability.

Health Assessment

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

Summary

Fix database error listeners to prevent Lambda crash

Adds error listeners to Neon and Node pools to stop unhandled exceptions that caused Lambda crashes, improving reliability for production deployments.

Health Assessment

• • Quick fix with minimal changes, AI-assisted code generation, and immediate merge after review.

Summary

Fix gateway client tool state and exit code output

Corrects lost plugin state in gateway client tool results and cleans up command output formatting, improving UI accuracy and reducing noise.

Health Assessment

• • Quick fix with minimal iterations, fast review, and small scope indicates low risk.

Summary

Fix Gemini usage conversion for zero token count

Corrects handling of zero cachedContentTokenCount in Gemini usage conversion, preventing errors and ensuring accurate token accounting.

Health Assessment

• • Fast review and minimal rework indicate a straightforward bug fix with clear test coverage.

Summary

Add copy session ID to topic dropdown

Adds a copy session ID action to the topic dropdown menu, enabling users to quickly copy the session ID to clipboard with a success toast.

Health Assessment

• • Fast cycle time of under an hour, minimal changes, AI-assisted development, and a single review comment indicate a smooth, low-risk PR.

Summary

Add home daily brief with linkable welcome and paired input hint

Introduces a per‑user daily brief surface on the home page, rotating welcome messages and hints to improve onboarding and user engagement.

Health Assessment

• • Merged within 1.8 hours with minimal review, indicating a straightforward implementation but a large code change.

Summary

Fix runtime session injection to use existing status

Corrects session handling to prevent race conditions and ensure consistent state across cloud runtime sessions, improving reliability.

Health Assessment

• • The PR was reviewed immediately and merged within an hour, indicating a smooth process with minimal friction.