Problem

The announce delivery mode for cron jobs has a hardcoded 60-second timeout cap in the codebase:

const waitMs = Math.min(params.announceTimeoutMs, 6e4)

Even if announceTimeoutMs is set higher (e.g. 180000), it's always capped to 60000ms.

Impact

Any isolated cron job that takes >60s to generate output will silently fail delivery. The job runs successfully, but the announce never reaches the user.

Error in journalctl:

Subagent announce failed: Error: gateway timeout after 60000ms
[cron:<jobId>] cron announce delivery failed

This affects common use cases like morning/evening digest jobs that aggregate data from multiple sources (typical duration: 70-180s).

Current Workaround

Set delivery.mode: "none" and add explicit message() tool call instructions in the cron prompt so the agent sends the message itself. Works but requires prompt engineering for every slow job.

Proposed Fix

Either:

1. Remove the cap and respect the configured announceTimeoutMs value
2. Raise the cap to something reasonable (300s / 5min)
3. Make the cap configurable via gateway config (e.g. cron.maxAnnounceTimeoutMs)

Option 1 is simplest — the user already explicitly sets announceTimeoutMs, so the cap is redundant.

Environment

• OpenClaw latest (npm)
• Multiple agents with digest cron jobs (60-180s duration)
• Related: Cron isolated jobs fail with 'Missing workspace template: AGENTS.md' despite file existing #14253 (AGENTS.md template bug in isolated cron)