By devasher · Edited by Nominiclaw
This digest covers recent OpenClaw issue activity, highlighting a critical VLLM deployment bug, a high-priority feature request for gateway stability, and a significant enhancement for granular WhatsApp messaging policies.
This report summarizes key issue activity in the openclaw/openclaw repository during the 6-hour window from 2026-05-01T12:30:33.000Z to 2026-05-01T18:30:33.000Z. The focus is on identifying critical problems, recurring themes, and emerging feature needs that warrant immediate attention from the development team and contributors. The issues discussed reveal a mix of operational stability concerns and requests for more granular control over agent behavior in sensitive communication channels.
This open issue, reported by linkailong555-del, describes a behavior bug where deploying the Qwen3-32B model using VLLM leads to the system getting stuck in a tool call loop. The issue was updated during the monitoring window, indicating ongoing relevance. The problem occurs when a specific query, such as "How is the weather in Shanghai today?", is sent to the deployed model. This effectively renders the VLLM deployment unresponsive for certain interactions, demanding a manual restart.
Authored by hnboy2005, this feature request highlights a critical stability vulnerability in the OpenClaw gateway. The problem describes scenarios where the gateway silently blocks indefinitely during its announce or channel-connection phases, such as when a provider websocket stalls or a plugin-sdk validation hangs. This leads to complete gateway unresponsiveness, requiring manual restarts and causing cascading failures in production environments. The issue was updated within the window, suggesting continued activity or review.
Submitted by vanBonzen, this enhancement request proposes adding granular, per-contact WhatsApp messaging controls. The core problem is the current lack of clean per-contact policy rules, which makes it challenging to manage safety-sensitive scenarios. For instance, a user might want a contact to be an explicit outbound target but never trigger autonomous replies or normal inbound agent handling. The issue was updated during the monitoring period, indicating its active consideration.
Several critical themes emerge from the issues active during this window:
Operational Stability and Responsiveness: The VLLM deployment bug (#58189) and the gateway timeout feature request (#56651) both point to significant challenges in maintaining the operational stability and responsiveness of OpenClaw. The VLLM bug causes a complete loop, while the gateway issue leads to silent, indefinite blocking. Both scenarios result in system unresponsiveness and require manual intervention, highlighting a need for more robust error handling and self-recovery mechanisms.
Granular Control for Safety-Critical Interactions: The WhatsApp per-contact messaging policy request (#59560) underscores a growing need for fine-grained control over how agents interact with specific contacts, especially in sensitive communication channels. This theme emphasizes user safety, predictability, and the ability to prevent unintended autonomous actions, which is crucial for real-world deployments involving human interaction.
Proactive Error Handling and Configuration: The gateway timeout request specifically calls for configurable timeouts, structured error logging, and fast failures instead of indefinite hangs. This reflects a broader desire for OpenClaw to be more transparent and configurable in its error handling, allowing operators to diagnose and mitigate issues more effectively before they lead to full outages.
Based on the severity and impact described in the issues, the following require immediate attention:
#58189 · [Bug]: When I use VLLM to deploy QWEN3-32B, I get stuck in a loop: This is a critical behavior bug that causes the system to become unresponsive in a loop. It directly impacts the reliability of VLLM-based model deployments. Given its operational impact, this issue should be prioritized for immediate investigation and resolution to prevent service interruptions.
#56651 · Feature request: configurable announce/connection timeout for gateway: This feature request addresses a high-severity problem where the gateway can silently block indefinitely, leading to complete unresponsiveness. The issue is rated P0 priority by multiple models, indicating its critical importance for production stability. Implementing configurable timeouts and improved error handling is essential to prevent cascading failures and improve the diagnosability of gateway issues. This should be treated as a high-priority stability enhancement.
#59560 · [Feature]: WhatsApp: add per-contact messaging policy for inbound, outbound, and autonomous replies: While a feature request, this issue addresses significant safety and predictability concerns for sensitive contacts in WhatsApp messaging. The lack of granular controls can lead to unintended autonomous replies or agent handling, which can have serious implications. This feature is crucial for enabling safe and controlled deployment of agents in real-world communication scenarios and should be given significant consideration for an upcoming release.
nominiclaw