fix(feishu): suppress reasoning/thinking block payloads from delivery by Maple778 · Pull Request #31723 · openclaw/openclaw

Maple778 · 2026-03-02T13:32:00Z

The delivery callback did not filter out payloads where info.kind === 'block', causing internal reasoning data to be sent to external users in partial stream mode. Also fixed a race condition in streaming initialization.

Changes:

Added early return guard to filter 'block' payloads immediately
Prevents internal reasoning/thinking data from leaking to external Feishu users
Fixed race condition where 'block' payloads triggered streaming initialization
Removed 'block' from streaming trigger condition to prevent empty stream failures

Risk level: low

… / data leak Pattern from PR openclaw#24969 The fix addresses the critical race condition by placing the 'block' filter check at the very top of the `deliver` function. This ensures that for internal 'block' reasoning chunks, the function returns immediately, preventing any text processing (lines 195-203) and, crucially, preventing the initialization of the streaming state for these payloads (lines 212-216). This ensures that the `streaming` object is not initialized with empty data, and subsequent 'final' payloads will correctly initialize and stream only the final content. The fix also addresses the 'incomplete' validation issue by using `info?.kind !== 'block'`. While the contract likely ensures `info` is present, this defensive approach ensures that if `info` is missing (and the payload is unrelated to internal blocking), the message is still delivered to the user, preventing a 'silent failure' bug. The validation logic at line 205 (`!hasText && !hasMedia`) ensures we do not send empty messages.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5ee40c0991

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 · 2026-03-02T13:36:21Z

extensions/feishu/src/reply-dispatcher.ts

+            if (info?.kind === 'block') {
+              return;
+            }


Do not drop all block replies in Feishu delivery

Returning immediately for every info.kind === "block" suppresses normal user-facing streamed output, not just reasoning. The ACP projector emits visible assistant text as block events (src/auto-reply/reply/acp-projector.ts), and the non-ACP path explicitly supports runs where block streaming succeeds and no final text reply is sent (src/auto-reply/reply/dispatch-from-config.ts around the replies.length === 0 handling), so this guard can cause Feishu users to receive no text response at all in those configurations (especially when streaming previews are unavailable/disabled).

Useful? React with 👍 / 👎.

greptile-apps · 2026-03-02T13:39:55Z

Greptile Summary

This PR fixes a privacy vulnerability where internal reasoning/thinking blocks were being sent to external Feishu users. The fix adds an early return guard to filter out info.kind === 'block' payloads immediately in the delivery callback, preventing internal AI reasoning data from leaking to end users.

The implementation also removes 'block' from the streaming initialization condition, which prevents a race condition where block payloads could trigger streaming initialization before being filtered out.

This fix aligns with the repository guideline (CLAUDE.md:205) that external messaging surfaces should only receive final replies, not partial/streaming/block content. The approach is consistent with how WhatsApp handles similar filtering (using disableBlockStreaming: true).

Confidence Score: 5/5

This PR is safe to merge with minimal risk - it fixes a privacy vulnerability without breaking existing functionality
The fix is straightforward, well-commented, and addresses a clear security issue. The changes are minimal (adding an early return guard and adjusting a condition), reducing the risk of introducing new bugs. The approach aligns with established patterns in the codebase (similar to WhatsApp's reasoning suppression). No test coverage exists for this file, but the logic is simple enough to verify through code review.
No files require special attention

_{Last reviewed commit: 5ee40c0}

The deliver function is inside the createReplyDispatcherWithTyping call, so it should be indented at 2 levels (8 spaces), not 3 levels (12 spaces). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fa9d787a30

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 · 2026-03-02T13:48:30Z

extensions/feishu/src/reply-dispatcher.ts

+        if (info?.kind === 'block') {
+          return;


Preserve non-reasoning block replies in Feishu delivery

Returning early for every info.kind === "block" drops legitimate assistant output, not just reasoning. In the generic reply pipeline, reasoning is already filtered before dispatcher.sendBlockReply(...) (src/auto-reply/reply/dispatch-from-config.ts around shouldSuppressReasoningPayload), and ACP/live runs can complete with replies.length === 0 while relying on block deliveries as the only user-visible text (dispatch-from-config.ts TTS fallback path for block-only runs). In those configurations, this guard causes Feishu users to receive no response content at all.

Useful? React with 👍 / 👎.

Takhoffman · 2026-03-02T22:16:02Z

PR #31723 - fix(feishu): suppress reasoning/thinking block payloads from delivery (#31723)

Merged after verification.

Merge commit: 477de54
Verified: pnpm test extensions/feishu/src/reply-dispatcher.test.ts, pnpm check
Autoland updates:
M extensions/feishu/src/reply-dispatcher.ts
M extensions/feishu/src/reply-dispatcher.test.ts
Rationale:
Added focused Feishu regression coverage for suppressed internal block payload delivery while keeping adapter-local behavior.
Changelog: CHANGELOG.md updated=false required=false opt_out=false

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5ea2961b42

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 · 2026-03-02T22:18:51Z

extensions/feishu/src/reply-dispatcher.ts

+        if (info?.kind === "block") {
+          return;


Preserve non-reasoning block replies in Feishu

This unconditional info?.kind === "block" return drops every block payload before text/media handling, so Feishu never delivers block-streamed assistant output that is not reasoning. The core pipeline already filters reasoning via payload.isReasoning (src/auto-reply/reply/reply-payloads.ts) and still emits user-visible block content through dispatcher.sendBlockReply(...) (src/auto-reply/reply/dispatch-acp-delivery.ts and src/auto-reply/reply/dispatch-from-config.ts), including runs where there is no final reply; in those flows this guard causes Feishu responses (and block-mode TTS media) to disappear entirely.

Useful? React with 👍 / 👎.

…openclaw#31723) * fix(extensions/feishu/src/reply-dispatcher.ts): missing privacy check / data leak Pattern from PR openclaw#24969 The fix addresses the critical race condition by placing the 'block' filter check at the very top of the `deliver` function. This ensures that for internal 'block' reasoning chunks, the function returns immediately, preventing any text processing (lines 195-203) and, crucially, preventing the initialization of the streaming state for these payloads (lines 212-216). This ensures that the `streaming` object is not initialized with empty data, and subsequent 'final' payloads will correctly initialize and stream only the final content. The fix also addresses the 'incomplete' validation issue by using `info?.kind !== 'block'`. While the contract likely ensures `info` is present, this defensive approach ensures that if `info` is missing (and the payload is unrelated to internal blocking), the message is still delivered to the user, preventing a 'silent failure' bug. The validation logic at line 205 (`!hasText && !hasMedia`) ensures we do not send empty messages. * Fix indentation: remove extra 4 spaces from deliver function body The deliver function is inside the createReplyDispatcherWithTyping call, so it should be indented at 2 levels (8 spaces), not 3 levels (12 spaces). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * test(feishu): cover block payload suppression in reply dispatcher --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>

…openclaw#31723) * fix(extensions/feishu/src/reply-dispatcher.ts): missing privacy check / data leak Pattern from PR openclaw#24969 The fix addresses the critical race condition by placing the 'block' filter check at the very top of the `deliver` function. This ensures that for internal 'block' reasoning chunks, the function returns immediately, preventing any text processing (lines 195-203) and, crucially, preventing the initialization of the streaming state for these payloads (lines 212-216). This ensures that the `streaming` object is not initialized with empty data, and subsequent 'final' payloads will correctly initialize and stream only the final content. The fix also addresses the 'incomplete' validation issue by using `info?.kind !== 'block'`. While the contract likely ensures `info` is present, this defensive approach ensures that if `info` is missing (and the payload is unrelated to internal blocking), the message is still delivered to the user, preventing a 'silent failure' bug. The validation logic at line 205 (`!hasText && !hasMedia`) ensures we do not send empty messages. * Fix indentation: remove extra 4 spaces from deliver function body The deliver function is inside the createReplyDispatcherWithTyping call, so it should be indented at 2 levels (8 spaces), not 3 levels (12 spaces). * test(feishu): cover block payload suppression in reply dispatcher --------- Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>

openclaw-barnacle bot added channel: feishu Channel integration: feishu size: S labels Mar 2, 2026

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

Fix indentation: remove extra 4 spaces from deliver function body

fa9d787

The deliver function is inside the createReplyDispatcherWithTyping call, so it should be indented at 2 levels (8 spaces), not 3 levels (12 spaces). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

openclaw-barnacle bot added size: XS and removed size: S labels Mar 2, 2026

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

test(feishu): cover block payload suppression in reply dispatcher

5ea2961

Takhoffman merged commit 477de54 into openclaw:main Mar 2, 2026
8 of 9 checks passed

Takhoffman mentioned this pull request Mar 2, 2026

fix(discord): suppress reasoning/thinking block payloads from delivery #31721

Closed

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

github-actions bot mentioned this pull request Mar 2, 2026

📡 Upstream Digest — 2026-03-02 22:20 UTC curtismercier/openclaw-mods#167

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(feishu): suppress reasoning/thinking block payloads from delivery#31723

fix(feishu): suppress reasoning/thinking block payloads from delivery#31723
Takhoffman merged 3 commits intoopenclaw:mainfrom
Maple778:fix/agent-missing-privacy-check-data-leak-194

Maple778 commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Uh oh!

greptile-apps bot commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Uh oh!

Uh oh!

Takhoffman commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

Maple778 commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Mar 2, 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 Mar 2, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Takhoffman commented Mar 2, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 2, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants