openclaw - 💡(How to fix) Fix [Bug]: WhatsApp channel silently drops long/complex agent responses (v2026.5.18) [3 comments, 3 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#84092Fetched 2026-05-20 03:44:10
View on GitHub
Comments
3
Participants
3
Timeline
16
Reactions
1
Author
Timeline (top)
labeled ×6commented ×3mentioned ×3subscribed ×3

Error Message

  1. Observe: the response never arrives on WhatsApp. No error is logged in gateway logs.

Root Cause

The 2026.5.18 release introduced a channel presentation capability negotiation system (changelog: "Plugins/messages: add presentation capability limits for channel renderers, adapt rich message controls before native rendering").

Telegram and Mattermost channels declare presentationCapabilityLimits in their channel-actions files (see related issues #56039, #55649, #56496). The WhatsApp channel plugin appears to be missing this capability declaration. When the core attempts to adapt a complex message for a renderer that returns no capabilities, it triggers a silent failure path instead of falling back to raw text delivery.

Fix Action

Workaround

Rollback to OpenClaw v2026.5.12:

npm install -g [email protected]

Code Example

npm install -g openclaw@2026.5.12
RAW_BUFFERClick to expand / collapse

Describe the Bug

After upgrading OpenClaw from 2026.5.12 to 2026.5.18, the WhatsApp channel silently drops long or complex agent responses (e.g., markdown tables, multi-line structured content over ~500 chars). Short responses (version checks, 1-line confirmations) deliver fine. Rolling back to 2026.5.12 restores full delivery.

To Reproduce

  1. Install OpenClaw v2026.5.18
  2. Configure WhatsApp channel (direct chat)
  3. Ask the agent a question that generates a response with markdown tables, code blocks, or ~1KB+ structured text
  4. Observe: the response never arrives on WhatsApp. No error is logged in gateway logs.
  5. Repeat with a short query ("what version?") — response arrives immediately.

Expected Behavior

All agent responses, regardless of length or formatting complexity, should be delivered to the WhatsApp channel as they did in v2026.5.12.

Environment

  • OpenClaw version: 2026.5.18 (regression), working on 2026.5.12
  • Platform: macOS (Mac Studio M4 Max, macOS 15.x)
  • WhatsApp plugin version: 2026.5.12 (pre-upgrade) and 2026.5.18 (updated — same broken behavior)
  • Node.js: v25.9.0
  • Channel: WhatsApp (direct chat, allowlist DM policy)
  • Config: textChunkLimit 100000, chunkMode "newline"

Root Cause Analysis

The 2026.5.18 release introduced a channel presentation capability negotiation system (changelog: "Plugins/messages: add presentation capability limits for channel renderers, adapt rich message controls before native rendering").

Telegram and Mattermost channels declare presentationCapabilityLimits in their channel-actions files (see related issues #56039, #55649, #56496). The WhatsApp channel plugin appears to be missing this capability declaration. When the core attempts to adapt a complex message for a renderer that returns no capabilities, it triggers a silent failure path instead of falling back to raw text delivery.

Suggested Fix

Add a presentationCapabilityLimits declaration to the WhatsApp channel plugin (analogous to the Telegram implementation in extensions/telegram/src/channel-actions.ts), declaring basic text markdown capability without rich controls (buttons, interactive elements). The adaptation layer should fall back to plain text on capability mismatch rather than silently dropping messages.

Workaround

Rollback to OpenClaw v2026.5.12:

npm install -g [email protected]

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…

Still need to ship something?

×6

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

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix [Bug]: WhatsApp channel silently drops long/complex agent responses (v2026.5.18) [3 comments, 3 participants]