Cron Digest with Native Notify

This playbook wires a timezone-aware cron job to Telegram, Slack, or a new web-console conversation. On a schedule, the agent builds a daily digest using shared conversation history, returns the digest as its final answer, and mono-agent delivers that final answer verbatim through native notification.

Who this is for

Data analysts and operators who want a scheduled briefing pushed to a chat without anyone asking for it — a morning digest that lands every day at a fixed local time.

Goal

A timezone-aware cron job that builds a daily digest with shared run history and delivers the final answer through native notification.

Features used

cron.scheduled-prompts — in-app scheduled prompts; see Cron. (config)
cron.jobs-dir — author jobs as cron/<id>.md frontmatter files; see Cron. (config)
channel.native-notify — notify: true delivers the cron final answer verbatim to Telegram, Slack, or exact web:new after a successful run; NOTHING_TO_REPORT (or empty final text) stays silent. See Delivery and Send Tools. (config)
slack.socket-mode or telegram.bot — the destination adapter that owns the allowlist. (config)
memory.journal — shared run history via conversationId; see Capture and Recall. (config)

Configuration

Enable the destination adapter and add the cron job. conversationId is the cron run-history thread; notifyConversationId is the delivery destination. Set it to exact web:new to create a separate CRON-marked web conversation for every distinct result; the web service must be running, and web is never inferred. If notifyConversationId is omitted, mono-agent only infers a destination when exactly one Telegram/Slack notify destination is available. With 0 or 2+ candidates delivery is skipped with a warning. Artifact-derived candidates use a 30-second in-process cache, measured from scan completion, to avoid rescanning a busy artifact directory for every notification. An artifact committed under a Telegram/Slack conversation id invalidates it at both running-summary and terminal-summary commits; runs using the default synthetic cron:/webhook: ids do not. Other artifact changes are picked up after cache expiry and the next scan completes. Cron model-exhaustion notices require an explicit notifyConversationId and never infer a destination. Put MONO_AGENT_SLACK_BOT_TOKEN and MONO_AGENT_SLACK_APP_TOKEN in .env; the source config omits credentials.

{
  "runtime": {
    "model": "anthropic:claude-sonnet-4-6"
  },
  "slack": {
    "enabled": true,
    "allowedChannelIds": ["C012345"],
    "unfurlLinks": false,
    "unfurlMedia": false
  },
  "cron": {
    "jobs": [
      {
        "id": "morning-digest",
        "enabled": true,
        "expression": "0 9 * * *",
        "timezone": "America/New_York",
        "prompt": "Build the morning digest. Your final answer is the digest to notify.",
        "conversationId": "daily-digest",
        "notify": true,
        "notifyConversationId": "slack:C012345"
      }
    ]
  }
}

Equivalent environment variables for the secrets and model:

export MONO_AGENT_MODEL="anthropic:claude-sonnet-4-6"
export MONO_AGENT_SLACK_BOT_TOKEN="xoxb-..."
export MONO_AGENT_SLACK_APP_TOKEN="xapp-..."

The same job can instead live in cron/morning-digest.md:

---
id: morning-digest
enabled: true
expression: "0 9 * * *"
timezone: America/New_York
conversationId: daily-digest
notify: true
notifyConversationId: slack:C012345
---

Build the morning digest. Your final answer is the digest to notify.

File jobs merge with cron.jobs; duplicate ids error.

The optional Slack unfurl settings above apply to the native cron notification because it uses the adapter’s normal message stream. Omit either key to preserve Slack’s current default for that preview type.

Steps

mono-agent init --model anthropic:claude-sonnet-4-6
Add the Telegram or Slack destination config and allowlist, or start mono-agent web for web:new.
Add the cron job with expression, conversationId, notify: true, and optionally notifyConversationId.
mono-agent validate, then mono-agent start.
Trigger a one-off tick or wait for the schedule and confirm the digest appears in the configured destination.