openclaw - 💡(How to fix) Fix WhatsApp typing indicator and message delivery appear delayed despite fast processing

Description

Messages sent via the WhatsApp channel appear quickly processed from OpenClaw's perspective (logs show immediate send), but the "typing..." indicator and actual message delivery/display on the WhatsApp client side experience a noticeable delay. This makes the agent appear slower than it actually is.

Steps to Reproduce

1. Configure WhatsApp channel with OpenClaw
2. Send a message to the agent via WhatsApp
3. Observe the time between:
   • Sending the message
   • The typing indicator appearing
   • The actual reply appearing

Expected Behavior

• Typing indicator should appear within ~500ms of message receipt
• Message should appear promptly after agent completes processing

Observed Behavior

• 1-3 second delay before typing indicator appears on WhatsApp client
• Even when the agent has already generated a response (visible in logs), the message takes additional time to display on the client
• The delay is primarily perceived on the WhatsApp client side — OpenClaw logs show processing and sending completing quickly

Investigation Root Causes

1. Pre-processing delay before typing starts: The typing indicator (typing.onReplyStart()) is called AFTER session loading, context building, and config validation. These operations can take 1-3 seconds before the first composing presence update is sent.

2. Redundant typing send in message function: sendMessageWhatsApp() in @openclaw/whatsapp/dist/send-D0iPQBMX.js calls sendComposingTo() right before sending the message. This duplicates the pipeline typing system and causes the typing indicator + message to arrive nearly simultaneously, making the typing indicator ineffective.

3. Coarse typing refresh interval: The typing keepalive refreshes every 3-6 seconds (hardcoded). For faster LLMs that respond in 5-10 sec, this means only 1-2 refreshes before the message arrives.

4. WhatsApp Desktop client-side rendering: The WhatsApp Desktop/Web client has known performance issues (Chromium wrapper, RAM usage), which adds to the perceived delay.

Proposed Fixes

1. Send typing indicator immediately upon message receipt — fire sendComposingTo() alongside the ack reaction, before heavy processing begins.

2. Remove redundant sendComposingTo() from sendMessageWhatsApp() (line 396 in send-D0iPQBMX.js) — the pipeline typing system already manages the typing lifecycle; this extra call is counterproductive (typing + message arrive simultaneously).

3. Reduce typing refresh interval / make it configurable — the 3-second keepalive should be adjustable, with lower default (1.5-2 seconds) for more responsive feel.

4. Add channels.whatsapp.immediateTypingIndicator config option — sends a composing presence update immediately when an inbound message is accepted.

Affected Components

• @openclaw/whatsapp plugin (send-D0iPQBMX.js, send-api-3hUuzYv1.js)
• Core reply pipeline (get-reply-DuA7xbHV.js, typing-C657onUT.js, reply-pipeline-C3CPjsQr.js)

Environment

• OpenClaw version: latest (May 2026)
• WhatsApp plugin: @openclaw/whatsapp
• OS: Linux (gateway host)