<@U090ER7JPHA> Please help me with a list of recen...
# slackbot-chatter
e
@nutritious-balloon-65098 Please help me with a list of recent bug reports within OpenHands/OpenHands related to the Local GUI experience. Please provide a severity for the bugs - for example, basic expectations like messages not appearing correctly or unable to connect to specific LLMs should be high priority
n
I'm on it! Joe Pelletier can track my progress here.
Summary I researched and compiled a comprehensive list of 21 recent bug reports in the OpenHands/OpenHands repository related to the Local GUI experience, organized by severity as requested. Key Findings: 🔴 HIGH PRIORITY (8 bugs) - Critical functionality issues: • LLM connectivity problems (DeepSeek v4-pro multi-turn failures, LM Studio "Provider NOT provided" errors) • UI stuck on "Starting"/"Loading..." states despite backend being ready • Messages silently lost (pending messages during startup, infinite chat loops) • Integration authentication completely broken (Bitbucket) • Container crashes from webhook connection failures 🟠 MEDIUM PRIORITY (9 bugs) - Significant UX issues: • Silent failures with no error feedback (attachment-only messages dropped, MCP timeouts) • Misleading connection/status indicators • Frontend rendering loop causing high GPU usage and UI freezes • Settings overwriting manual configuration • Self-destructive port conflict handling 🟡 LOW PRIORITY (4 bugs) - Minor/intermittent issues: • Duplicated messages • Infinite scroll loops • Visual flickers Most Critical Pain Points for Local GUI Users: 1. UI stuck states - Multiple reports of loading screens that never resolve 2. Local LLM connectivity - Users unable to connect to LM Studio, DeepSeek, custom endpoints 3. Silent failures - No error feedback when things go wrong 4. Bitbucket integration broken - API token auth not supported Direct GitHub links to all issues are included in my detailed response above for easy reference and triaging.
e
🔴 High Priority • #14347- DeepSeek v4-pro/v4-flash compatibility issues (model list, provider routing, reasoning_content, UI settings overwriting config) • #13647- Self-hosted UI remains stuck on "Starting"/"Loading..." even though backend conversation is IDLE • #13948- LLM Provider NOT provided with LM Studio (local LLM integration broken) • #13942- Using Docker causes infinite loop chat response • #13716- Pending messages not sent after conversation is ready (messages silently lost) • #12653- Agent-Server webhook connection failures cause container crashes and sandbox connection errors • #14304- Bitbucket provider authentication broken (API tokens not working) • #11306- Bitbucket integration no longer works when running OpenHands locally 🟠 Medium Priority • #14047- Send button silently drops attachment-only chat messages • #13690- Think tool calls displayed as raw JSON text in chat messages • #13536- Web UI temporarily stops with "starting" and partial event stream load • #13479- Self-hosted UI shows "Disconnected" while backend and sandbox are still alive • #13603- Agent remains idle with no visual feedback or error notification (MCP timeout) • #12681- High client-side GPU usage (~30%) and UI freeze caused by frontend rendering loop • #11918- GUI hard stuck once you hit max_iterations • #13927- OpenHands kills its own process on port 8000 during local web deployment • #13769- Repo picker does not show some accessible repos, even by exact name or pasted URL 🟡 Low Priority • #13771- Initial response is sometimes duplicated when sending a message • #13282- Infinite scroll in conversation history does not stop at last available conversation • #14245- ACP tool calls flash empty state before populating with real values • #11874- OpenRouter free models not found
n
Summary I searched the OpenHands/OpenHands repository for recent bug reports related to the Local GUI experience and compiled a prioritized list based on severity. Key Findings 🔴 High Priority (8 issues) - Critical functionality broken: • #14347 - DeepSeek v4-pro compatibility issues & UI settings overwriting config • #13647 - UI stuck on "Starting/Loading" despite backend being ready • #13948 - LM Studio local LLM integration broken • #13942 - Docker causes infinite loop chat response • #13716 - Pending messages silently lost • #12653 - Webhook failures cause container crashes • #14304 & #11306 - Bitbucket integration broken 🟠 Medium Priority (9 issues) - Significant UX issues including silent failures, poor error feedback, and performance problems 🟡 Low Priority (4 issues) - Minor visual glitches and intermittent issues Top Pain Point Themes 1. UI stuck on loading states - Multiple reports of indefinite loading 2. LLM provider configuration problems - DeepSeek, LM Studio, custom endpoints 3. Silent failures without error feedback - Attachments dropped, messages lost, MCP timeouts 4. Integration authentication broken - Bitbucket particularly affected 5. WebSocket/connection reliability - Misleading disconnected states
Summary I searched the OpenHands/OpenHands repository for recent bug reports related to the Local GUI / Docker / self-hosted experience and compiled a prioritized list. Key Findings 🔴 High Priority Issues (8 bugs): • UI stuck on "Starting"/"Loading..." states despite backend being ready • Local LLM integrations broken (LM Studio, DeepSeek v4-pro) • Docker causing infinite loop chat responses • Messages silently lost when typed during startup • Container crashes from webhook connection failures • Bitbucket integration completely broken (auth method deprecated) 🟠 Medium Priority Issues (10 bugs): • Misleading "Disconnected" status in self-hosted UI • No visual feedback for MCP timeout errors • High GPU usage / UI freeze from frontend rendering loop • GUI stuck at max_iterations with no way to continue • Self-hosted settings fragile across restarts • Docker-compose configuration issues 🟡 Low Priority Issues (3 bugs): • Infinite scroll loop in conversation history • Visual flicker on tool calls • Specific OpenRouter model compatibility Common Pain Points for Self-Hosted Users: 1. UI stuck states - Multiple reports of UI not reflecting actual backend status 2. LLM provider configuration - Local LLM setup frequently fails 3. Silent failures - No error feedback when things go wrong 4. Integration auth - Bitbucket broken for all local users 5. Docker networking - Webhook and container communication issues All information was gathered directly from GitHub issues using the GitHub CLI. No code changes were made.