Step-by-Step Fix
1. Reproduce the Bug and Capture the Details
Before reporting, try to reproduce the bug at least twice. Note whether it happens:
- Every time (deterministic) or only sometimes (intermittent)
- Only with specific agent configurations or with any agent
- Only in certain environments (specific browser, OS, or network)
Intermittent bugs are harder to fix, so documenting the frequency and conditions is especially important.
2. Collect the Run ID
Open the OpenClaw dashboard and navigate to the failed run. The run ID is in the page URL:
https://app.openclaw.com/runs/run_abc123xyz456
Copy run_abc123xyz456 — this is what you need. The run ID allows the support team to access internal execution logs, the exact tool call sequence, error stack traces, and infrastructure state that are not exposed in the user-facing log view.
3. Copy the Exact Error Message
Do not paraphrase error messages. Copy the exact text, including:
- HTTP status codes (401, 429, 500, etc.)
- Error type fields (
authentication_error,rate_limit_error, etc.) - Any request IDs embedded in the error response
- Stack traces if visible in the log
If the error occurred in the browser UI rather than a run log, open your browser's developer tools (F12), go to the Console tab, and copy any red error messages there.
4. Prepare Your Agent Configuration
Export or copy your agent configuration file. Before including it in the bug report, scrub all sensitive values:
# Original (DO NOT share):
anthropic_api_key: sk-ant-abc123def456
# Sanitized version (safe to share):
anthropic_api_key: REDACTED
Keep the structure, model names, tool configurations, and timeout settings intact — these are often the source of the bug.
5. Write the Bug Report
Use this template:
**Summary**: [One sentence: what broke]
**Expected behavior**: [What should have happened]
**Actual behavior**: [What actually happened]
**Error message**:
[Paste exact error text here]
**Steps to reproduce**:
1. [First action]
2. [Second action]
3. [etc.]
**Run ID**: run_abc123xyz456
**Plan tier**: [Free / Pro / Team / Enterprise]
**Date and time**: [When the failure occurred, with timezone]
**Regression?**: [Did this work before? If yes, when did it break?]
**Agent config** (API keys removed):
[Paste sanitized YAML here]
6. Submit Via the Appropriate Channel
- Billing or account issues: Use the support form at openclaw.com — these require private handling
- Technical bugs with run IDs: Post in #bug-reports on OpenClaw's Discord for fastest response
- Production-critical issues: Use both channels and add "PRODUCTION IMPACT" to the subject
Follow up if you do not receive a response within 24 hours. Reference your original ticket or Discord message.
Why This Happens
Bugs in OpenClaw typically fall into three categories: platform bugs in the agent execution engine (rare, fixed by the OpenClaw team), configuration issues that produce unexpected behavior (most common, often resolvable by the user), and Anthropic API changes that break existing agent behaviors (periodic, usually announced via Anthropic's changelog). A good bug report helps the support team distinguish between these categories in minutes instead of hours.
Common Mistakes to Avoid
- Saying "it doesn't work" without specifics: A vague report gets a generic response. "The browser tool returns a 503 error when navigating to URLs with HTTPS redirects" is actionable; "the browser doesn't work" is not.
- Including actual API keys in reports: Always redact credentials before submitting. Even in private support tickets, minimizing exposure is good practice.
- Reporting without attempting to reproduce first: If you cannot reproduce it yourself, say so explicitly and include the timestamp so the team can correlate it with server-side logs.
- Submitting duplicate reports across multiple channels without noting it: If you post in Discord and also submit a ticket, mention it ("I also filed support ticket #12345") so the team does not duplicate effort.
FAQ
Q: How long does OpenClaw support typically take to respond?
OpenClaw support typically responds to tickets submitted through the support form at openclaw.com within 24 to 48 business hours. Discord responses in the #bug-reports channel are often faster — the community can confirm reproduction within hours, and OpenClaw engineers monitor the channel during business hours. For production-critical issues, label your report clearly with "PRODUCTION IMPACT" in both the ticket subject and Discord message to get priority attention.
Q: Should I report a bug if I can work around it?
Yes, even if you have a workaround. Workarounds mask bugs that may affect other users who have not found the same workaround. Report the bug with your workaround included — the workaround helps the team understand the scope of the issue and may speed up diagnosis. Include the workaround in the "Actual behavior" section so support knows what you tried.
Q: What if my bug involves sensitive data I cannot share?
Use the private support form at openclaw.com rather than the public Discord channel. In the report, describe the data type ("the bug occurs when the browser tool fetches a page requiring authentication") without including actual credentials or data. The run ID is the most valuable piece of information — it lets support inspect your execution internally without you needing to share the sensitive data directly.
Q: How do I know if my bug has already been reported?
Search the #bug-reports Discord channel before posting. Use the search function with keywords from your error message (the HTTP status code, the tool name, or the specific error type). If you find a matching report, add a reply confirming you can reproduce it and include your run ID — this helps the team understand the scope and prioritize the fix.
Related Issues
- OpenClaw cron job failing – find logs and root cause
- OpenClaw tool calls failing – timeouts and not available
- OpenClaw agent not responding – stuck run
- OpenClaw 401 invalid API key errors in cron jobs
Additional FAQ
Q: How do I know if the problem is on my end or the platform's side? Check the platform's official status page first — most services maintain a public status page that shows current incidents and outages. If no incident is posted and the problem only affects your account (not reported widely on Reddit or Twitter), it is likely a local issue. Testing in incognito mode and on a different network also helps isolate whether the problem is browser-specific, network-specific, or account-specific.