fix(openrouter): handle reasoning_details in Qwen3 stream parsing by bladin · Pull Request #66905 · openclaw/openclaw

bladin · 2026-04-15T01:19:54Z

Summary

Fix OpenRouter/Qwen3 models returning empty responses due to unrecognized reasoning_details field.

Problem: When using openrouter/qwen/qwen3-235b-a22b or any Qwen3 model via OpenRouter, every response fails with "incomplete turn detected: payloads=0" and the user sees "⚠️ Agent couldn't generate a response."

Root Cause: The stream parser in src/agents/openai-transport-stream.ts handled reasoning_content, reasoning, and reasoning_text but not reasoning_details — which Qwen3 returns via OpenRouter. The unhandled field caused zero content blocks to be assembled, triggering the incomplete turn path.

Solution: Add handling for reasoning_details array. Extract text from items with type: "reasoning.text" and treat as thinking content, enabling successful turn completion.

Additional defensive fixes:

Add ?? "" guards in src/channels/plugins/setup-wizard.ts:458 and src/channels/plugins/setup-wizard-helpers.ts:994 to prevent potential undefined trim errors from clack wrapper edge cases (belt-and-suspenders approach mentioned in issue comments).

Changes

openai-transport-stream.ts: Add reasoning_details handling in processOpenAICompletionsStream, extracting reasoning text from array items and creating thinking blocks.
openai-transport-stream.test.ts: Add test case verifying correct handling of Qwen3 reasoning_details.
setup-wizard.ts & setup-wizard-helpers.ts: Add defensive guards around prompter.text() results.

Test plan

New unit test: handles reasoning_details from OpenRouter/Qwen3 in completions stream passes
All 53 existing openai-transport-stream.test.ts tests pass
Fix verified by community members in original issue thread

AI Code Generation Disclosure

🤖 This PR was created with AI assistance (Claude Codex)
The fix implementation and test were generated based on issue description and community diagnosis

Environment

OpenClaw version: 2026.4.14 (base)
Tested with: Bun test runner, TypeScript type-checking
Provider: OpenRouter
Model: openrouter/qwen/qwen3-235b-a22b, qwen3-32b

Checklist

Build successful (no type errors)
Tests added/updated and passing
Issue reference added (Fixes [Bug]: OpenRouter/Qwen3 stream parsing fails — reasoning_details field not handled, causes payloads=0 #66833)
Commit message follows conventional format
PR description detailed with test plan
Defensive coding practices applied

greptile-apps · 2026-04-15T01:23:37Z

Greptile Summary

This PR fixes empty responses from OpenRouter/Qwen3 models by adding reasoning_details array handling to the completions stream parser, following the same pattern used for reasoning_content/reasoning/reasoning_text. It also adds ?? "" guards in two setup-wizard prompt calls to prevent potential TypeError when the prompter returns null/undefined.

The implementation is targeted and consistent with existing reasoning-field handling. The new test covers the primary multi-chunk scenario and confirms correct assembly of thinking + text content blocks.

Confidence Score: 5/5

Safe to merge — fix is minimal, follows existing patterns, and is covered by a new unit test.

No P0 or P1 issues found. The reasoning_details handling mirrors the established pattern for other reasoning fields. The thinkingSignature: "reasoning_details" value is consistent with how e.g. "reasoning_content" is stored for the same field, and any cross-transport concern (JSON.parse on history replay) is pre-existing and not introduced here. Defensive ?? "" guards are safe.

No files require special attention.

_{Reviews (1): Last reviewed commit: "fix(openrouter): handle reasoning_detail..." | Re-trigger Greptile}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 94cda6c34f

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-15T01:26:11Z

@@ -455,7 +455,7 @@ export function buildChannelSetupWizardAdapterFromSetupWizard(params: {
              });
            },
          });
-          const trimmedValue = rawValue.trim();
+          const trimmedValue = (rawValue ?? "").trim();


Preserve required-input failure on undefined text responses

Coalescing rawValue to "" here turns an undefined prompt result into a silent empty value. In the required-field path, that empty value falls through to normalizedValue being falsy and the loop continue branch, so setup can proceed without collecting a required input. This is most visible when the prompt layer returns undefined on cancel/interruption: instead of failing fast, the wizard can now silently skip mandatory configuration.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-15T01:26:11Z

+      (await params.prompter.text({
        message: params.inputPrompt,
        validate: (value) => (value?.trim() ? undefined : "Required"),
-      })
+      })) ?? ""
    ).trim();


Keep token prompts from downgrading undefined to "keep"

Defaulting prompter.text() to "" causes interrupted/undefined token entry to be treated as an empty token rather than an error. Downstream, empty token values are interpreted as "keep existing token", which is incorrect for first-time setup (no configured token) and can leave the credential unset without any explicit failure or retry. This converts a hard failure into silent misconfiguration.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4fb723f568

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: bcfac3bfe7

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

Add support for the reasoning_details field returned by OpenRouter/Qwen3 models. Previously this field was not recognized, causing payloads=0 and incomplete turn errors. - Add reasoning_details handling in processOpenAICompletionsStream - Extract text from reasoning_details array items with type reasoning.text - Treat as thinking content, similar to other reasoning fields - Add test case for reasoning_details handling Fixes openclaw#66833

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 501561be5f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

@bladin

… (thanks @bladin)

@bladin

…) (thanks @bladin)

obviyus

Verified the OpenRouter Qwen3 empty-reply failure path and confirmed the stream parser now handles reasoning_details on the main completions path.

Maintainer follow-up: preserved same-chunk tool calls, kept streamed tool-call argument accumulation intact across repeated reasoning_details deltas, added regression coverage for both cases, and added the unreleased changelog entry.

Local gate: pnpm test src/agents/openai-transport-stream.test.ts (55/55). pnpm check was rerun and only hit the unrelated current-main src/media-understanding/runner.proxy.test.ts:231 tsgo failure.

obviyus · 2026-04-15T02:46:23Z

Landed on main.

Landed source commit: 406dbfc
Merge commit: e0bf756

Thanks @bladin.

@bladin

… (thanks @bladin) * fix(openrouter): handle reasoning_details field in Qwen3 stream parsing Add support for the reasoning_details field returned by OpenRouter/Qwen3 models. Previously this field was not recognized, causing payloads=0 and incomplete turn errors. - Add reasoning_details handling in processOpenAICompletionsStream - Extract text from reasoning_details array items with type reasoning.text - Treat as thinking content, similar to other reasoning fields - Add test case for reasoning_details handling Fixes openclaw#66833 * fix(openrouter): keep tool calls with reasoning_details * fix: handle OpenRouter Qwen3 reasoning_details streams (openclaw#66905) (thanks @bladin) * fix: preserve streamed tool calls with reasoning deltas (openclaw#66905) (thanks @bladin) --------- Co-authored-by: bladin <[email protected]> Co-authored-by: Ayaan Zaidi <[email protected]>

@bladin

… (thanks @bladin) * fix(openrouter): handle reasoning_details field in Qwen3 stream parsing Add support for the reasoning_details field returned by OpenRouter/Qwen3 models. Previously this field was not recognized, causing payloads=0 and incomplete turn errors. - Add reasoning_details handling in processOpenAICompletionsStream - Extract text from reasoning_details array items with type reasoning.text - Treat as thinking content, similar to other reasoning fields - Add test case for reasoning_details handling Fixes openclaw#66833 * fix(openrouter): keep tool calls with reasoning_details * fix: handle OpenRouter Qwen3 reasoning_details streams (openclaw#66905) (thanks @bladin) * fix: preserve streamed tool calls with reasoning deltas (openclaw#66905) (thanks @bladin) --------- Co-authored-by: bladin <[email protected]> Co-authored-by: Ayaan Zaidi <[email protected]>

@bladin

… (thanks @bladin) * fix(openrouter): handle reasoning_details field in Qwen3 stream parsing Add support for the reasoning_details field returned by OpenRouter/Qwen3 models. Previously this field was not recognized, causing payloads=0 and incomplete turn errors. - Add reasoning_details handling in processOpenAICompletionsStream - Extract text from reasoning_details array items with type reasoning.text - Treat as thinking content, similar to other reasoning fields - Add test case for reasoning_details handling Fixes openclaw#66833 * fix(openrouter): keep tool calls with reasoning_details * fix: handle OpenRouter Qwen3 reasoning_details streams (openclaw#66905) (thanks @bladin) * fix: preserve streamed tool calls with reasoning deltas (openclaw#66905) (thanks @bladin) --------- Co-authored-by: bladin <[email protected]> Co-authored-by: Ayaan Zaidi <[email protected]>

@bladin

… (thanks @bladin) * fix(openrouter): handle reasoning_details field in Qwen3 stream parsing Add support for the reasoning_details field returned by OpenRouter/Qwen3 models. Previously this field was not recognized, causing payloads=0 and incomplete turn errors. - Add reasoning_details handling in processOpenAICompletionsStream - Extract text from reasoning_details array items with type reasoning.text - Treat as thinking content, similar to other reasoning fields - Add test case for reasoning_details handling Fixes openclaw#66833 * fix(openrouter): keep tool calls with reasoning_details * fix: handle OpenRouter Qwen3 reasoning_details streams (openclaw#66905) (thanks @bladin) * fix: preserve streamed tool calls with reasoning deltas (openclaw#66905) (thanks @bladin) --------- Co-authored-by: bladin <[email protected]> Co-authored-by: Ayaan Zaidi <[email protected]>

openclaw-barnacle Bot added agents Agent runtime and tooling size: S labels Apr 15, 2026

chatgpt-codex-connector Bot reviewed Apr 15, 2026

View reviewed changes

bladin force-pushed the fix/66833-openrouter-qwen3-reasoning-details branch from 94cda6c to 4fb723f Compare April 15, 2026 01:31

chatgpt-codex-connector Bot reviewed Apr 15, 2026

View reviewed changes

Comment thread src/agents/openai-transport-stream.ts Outdated

chatgpt-codex-connector Bot reviewed Apr 15, 2026

View reviewed changes

Comment thread src/agents/openai-transport-stream.ts Outdated

openclaw-barnacle Bot added size: M and removed size: S labels Apr 15, 2026

obviyus self-assigned this Apr 15, 2026

bladin and others added 2 commits April 15, 2026 08:07

fix(openrouter): keep tool calls with reasoning_details

1ba15da

chatgpt-codex-connector Bot reviewed Apr 15, 2026

View reviewed changes

Comment thread src/agents/openai-transport-stream.ts Outdated

fix: handle OpenRouter Qwen3 reasoning_details streams (openclaw#66905)…

35a2bc7

… (thanks @bladin)

obviyus force-pushed the fix/66833-openrouter-qwen3-reasoning-details branch from 501561b to 35a2bc7 Compare April 15, 2026 02:42

fix: preserve streamed tool calls with reasoning deltas (openclaw#66905…

406dbfc

…) (thanks @bladin)

obviyus approved these changes Apr 15, 2026

View reviewed changes

obviyus merged commit e0bf756 into openclaw:main Apr 15, 2026
9 checks passed

github-actions Bot mentioned this pull request Apr 15, 2026

📡 Upstream Digest — 2026-04-15 04:40 UTC curtismercier/openclaw-mods#577

Open

Magicray1217 mentioned this pull request Apr 15, 2026

[Bug]: OpenRouter/Qwen3 stream parsing fails — reasoning_details field not handled, causes payloads=0 #66833

Closed

This was referenced Apr 15, 2026

[Bug]: Empty completed turns in ghcr.io/openclaw/openclaw:2026.4.14 #66768

Closed

OpenRouter models returning reasoning_details get content silently dropped (minimax-m2.7, not just Qwen3) #67410

Closed

briandevans mentioned this pull request Apr 16, 2026

fix(agents): handle OpenRouter reasoning_details in stream parser #66842

Closed

25 tasks

neeravmakwana mentioned this pull request Apr 18, 2026

fix(openrouter): surface reasoning_details visible-text as user output (#68261) #68351

Closed

20 tasks

vincentkoc mentioned this pull request Apr 18, 2026

fix(openrouter): parse visible reasoning_details output #68577

Merged

25 tasks

clawsweeper Bot mentioned this pull request May 5, 2026

fix(agents): normalize structured delta.content blocks to prevent [object Object] in chat replies #75339

Open

5 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(openrouter): handle reasoning_details in Qwen3 stream parsing#66905

fix(openrouter): handle reasoning_details in Qwen3 stream parsing#66905
obviyus merged 4 commits intoopenclaw:mainfrom
bladin:fix/66833-openrouter-qwen3-reasoning-details

bladin commented Apr 15, 2026

Uh oh!

greptile-apps Bot commented Apr 15, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Apr 15, 2026

Uh oh!

chatgpt-codex-connector Bot Apr 15, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

obviyus left a comment

Uh oh!

Uh oh!

obviyus commented Apr 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

bladin commented Apr 15, 2026

Summary

Changes

Test plan

AI Code Generation Disclosure

Environment

Checklist

Uh oh!

greptile-apps Bot commented Apr 15, 2026

Greptile Summary

Confidence Score: 5/5

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

obviyus left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

obviyus commented Apr 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants