Issue Station
Sign In
Tools hereToolbox

openclaw - ✅(Solved) Fix Auth probe sends max_tokens: 1 which fails on OpenAI models with min 16 [1 pull requests, 1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#65703•Fetched 2026-04-14 05:40:47
View on GitHub
Comments
0
Participants
1
Timeline
2
Reactions
0
Author
abnershangabnershang
Participants
abnershangabnershang
Timeline (top)
cross-referenced ×1referenced ×1

Fix Action

Workaround

Custom curl-based probe with max_tokens: 16.

PR fix notes

PR #65738: fix(onboard): use max_tokens=16 for OpenAI-compat verification probes (Closes #65703)

Description (problem / solution / changelog)

Summary

The chat-completions verification probe in `requestOpenAICompatVerification` (`src/commands/onboard-custom.ts`) sent `max_tokens: 1`, which fails against OpenAI — and compatible proxies routing to recent OpenAI models — because they enforce `max_tokens >= 16`. The failure surfaces as `format_error` on `openclaw models status --probe` for any `zenmux-openai-chat/*` (or similar) model (#65703).

The Azure Responses API path already uses `max_output_tokens: 16`, so there's precedent for a higher minimum. This PR raises the chat-completions path to match.

The Anthropic probe is left alone — Anthropic accepts `max_tokens: 1`.

Closes #65703.

Changes

  • `src/commands/onboard-custom.ts`: chat-completions probe now sends `max_tokens: 16` with a short comment explaining why
  • `src/commands/onboard-custom.test.ts`: update the OpenAI probe test and the Azure Foundry chat-completions test to expect `max_tokens: 16`; leave the Anthropic test at `max_tokens: 1`

Test plan

  • Two affected tests updated to match new expected value
  • Anthropic probe tests untouched
  • Minimal diff, no behavior change for non-probe paths

🤖 Generated with Claude Code

Changed files

  • src/commands/onboard-custom.test.ts (modified, +2/-2)
  • src/commands/onboard-custom.ts (modified, +4/-1)

Code Example

openclaw models status --probe

---

format · 2.7s
↳ LLM request failed: provider rejected the request schema or tool payload.
RAW_BUFFERClick to expand / collapse

Bug

openclaw models status --probe sends max_tokens: 1 for OpenAI-compatible endpoints (onboard-KJNA-AL1.js line ~212), but OpenAI models enforce max_output_tokens >= 16. This causes all OpenAI probes to return format_error.

The Vertex AI probe path already uses max_output_tokens: 16 (line ~200), so there's precedent for a higher minimum.

Steps to reproduce

openclaw models status --probe

Any zenmux-openai-chat/* model will show:

format · 2.7s
↳ LLM request failed: provider rejected the request schema or tool payload.

Expected

Probe should use max_tokens: 16 (or higher) to satisfy the upstream minimum.

Workaround

Custom curl-based probe with max_tokens: 16.

extent analysis

TL;DR

Update the openclaw models status --probe command to use max_tokens: 16 or higher for OpenAI-compatible endpoints.

Guidance

  • Review the onboard-KJNA-AL1.js file (line ~212) to verify the current max_tokens value being sent.
  • Consider updating the max_tokens value to 16 or higher to match the Vertex AI probe path precedent (line ~200).
  • Test the updated openclaw models status --probe command with the new max_tokens value to ensure successful probe execution.
  • If issues persist, use the custom curl-based probe with max_tokens: 16 as a temporary workaround.

Example

No code snippet is provided as the issue implies a configuration or parameter update rather than a code change.

Notes

The solution assumes that updating the max_tokens value will resolve the format_error issue with OpenAI models. However, this may not be the only factor contributing to the error.

Recommendation

Apply workaround: use max_tokens: 16 or higher for OpenAI-compatible endpoints, as this change is consistent with the existing Vertex AI probe path configuration and is likely to resolve the format_error issue.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#agent execution#callback error#memory management#API rate limit#retriever error

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING