Known Issues

This page tracks known issues affecting Hosted Claws instances and their workarounds.

Issue	Status	Affected
Config Editor Slow to Load	Awaiting upstream fix	All instances
Tool Error Warnings After Successful Operations	Awaiting upstream fix	All instances

Config Editor Slow to Load

Status: Awaiting upstream fix
Affected: All instances
Upstream Issue: openclaw/openclaw#10384

Symptoms

When opening the Config Editor in the WebUI for the first time, the page may appear to hang or load indefinitely for 2-3 minutes. The gateway remains responsive during this time, but the WebUI tab will appear stuck.

Cause

The Config Editor requests the full OpenClaw configuration schema on first load. This schema is large (~400KB) and takes significant time to generate and transmit on resource-constrained containers.

Workarounds

Option 1: Wait for initial load

The first load takes 2-3 minutes, but the schema is then cached in your browser. Subsequent visits to the Config Editor will load instantly (until you clear browser data).

Option 2: Use chat commands instead

If you have a messaging channel connected (Discord, Telegram, etc.), you can manage your configuration entirely via chat commands:

/config get models.default          # View a setting
/config set models.default gpt-5    # Change a setting
/config list                        # See available options

This bypasses the WebUI entirely and responds immediately.

Resolution

We are monitoring the upstream issue and will update all customer instances as soon as a fix is released.

Tool Error Warnings After Successful Operations

Status: Awaiting upstream fix
Affected: All instances
Upstream Issue: openclaw/openclaw#39406

Symptoms

You may occasionally see warning messages like:

⚠️ 📝 Edit: in ~/IDENTITY.md (521 chars) failed

...even when your assistant reports that the task completed successfully.

Cause

When a tool (like file edit) fails on the first attempt, OpenClaw immediately shows a warning. If the assistant retries and succeeds, you'll see both:

The success message ("Done!" or similar)
The earlier warning bubble

This is intentional upstream behavior — OpenClaw surfaces all tool errors, even transient ones the assistant recovered from.

Workarounds

Option 1: Trust the final message

If your assistant says the task succeeded, it did. The warning refers to an earlier failed attempt that was automatically retried.

Option 2: Check for confirmation

Ask your assistant: "Did that actually work?" — it will confirm whether the operation succeeded.

What the warnings mean

Warning	What happened
`Edit: ... failed`	File edit content didn't match; assistant likely retried with corrected content
`exec: ... failed`	Command failed; assistant may have adjusted and retried
`read: ... failed`	File not found or inaccessible; assistant may have tried an alternative path

Resolution

We've raised a feature request to add a config option that suppresses transient errors when the assistant self-corrects. We'll update all instances when this is available.

Last updated: 8 March 2026