…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

… finally block

PR openclaw#26881 fixed the followup runner's missing typing.markDispatchIdle()
call, but the same gap existed in the main reply pipeline (runReplyAgent).
The main pipeline relied solely on the buffered dispatcher's finally block
(dispatch.ts) to fire markDispatchIdle(). When the reply path exits before
that finally executes — early error, abort, or code path that bypasses the
dispatcher — the typing keepalive loop spins indefinitely (same root cause
as openclaw#26993, openclaw#27053).

Fix: add typing.markDispatchIdle() as a safety-net in runReplyAgent's own
finally block, immediately after typing.markRunComplete().  Calling it
twice is harmless: maybeStopOnIdle() is guarded by both the runComplete
and dispatchIdle flags, so the second invocation is a no-op.

Adds two regression tests to agent-runner.runreplyagent.test.ts:
  - markDispatchIdle is fired on successful run
  - markDispatchIdle is fired even when the agent throws

Fixes openclaw#27172
Related: openclaw#26881 (same pattern, followup runner)

Co-Authored-By: Claude <noreply@anthropic.com>

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

(cherry picked from commit 5e1bfb2)

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

(cherry picked from commit 5e1bfb2)

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

Restores code that was accidentally removed alongside the continuation
feature in the initial commit (e0dc060). These are upstream safety
mechanisms that the feature lich didn't understand:

1. catch(error) block in runReplyAgent — keeps followup queue moving
   when an unexpected exception escapes the run path
2. typing.markDispatchIdle() in finally — stops typing keepalive on
   early exit/error (fix for openclaw#26881)
3. hasSessionRelatedCronJobs check — suppresses false 'I didn't schedule
   a reminder' warnings when existing cron job covers the commitment
   (fix for openclaw#32228)

These regressions were correctly identified by automated review
(chatgpt-codex-connector[bot] and greptile-apps[bot]).

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>

…k indicator (openclaw#26881)

The followup runner (used for queued messages, inter-agent sends,
heartbeat followups, etc.) only called typing.markRunComplete() in
its finally block.  The typing controller requires BOTH markRunComplete
AND markDispatchIdle to trigger cleanup — but markDispatchIdle was
only wired through the buffered dispatcher path, which followup turns
bypass entirely.

This caused the typing indicator to persist indefinitely on channels
like Telegram when the agent replied with NO_REPLY or produced empty
payloads, because the keepalive loop was never stopped.

Adds markDispatchIdle() alongside markRunComplete() in the followup
runner's finally block, and four test cases covering NO_REPLY, empty
payloads, agent errors, and successful delivery.

Complements openclaw#26295 which addressed the channel-level callback layer.

Fixes openclaw#26595

Co-authored-by: Samantha <samantha@Samanthas-Mac-mini.local>