stall-recovery

name	stall-recovery
description	Execute a stall deal check — query stalled deals, send follow-ups per stage policy, archive long-inactive deals as Lost.

Stall Recovery

When triggered for stall deal check:

1. Read /sales-pipeline/stall-deals-policy — get stall_days, archive_days, outreach_cutoff_stage_id, stage_actions.
2. Compute stall_ts = now - stall_days x 86400. Query only stalled deals — never fetch all open deals. Also filter to relevant pipelines and non-WON/LOST stages at query time if the policy lists specific pipeline_ids:

erp_table_data(table_name="crm_deal", options={
    "filters": {"AND": [
        "contact.contact_last_outbound_ts:<:STALL_TS",
        "contact.contact_last_inbound_ts:<:STALL_TS",
    ]},
    "include": ["contact"],
})

3. Per deal, look up its stage in stage_actions:
   • skip: stage is below outreach cutoff — do nothing
   • email: send follow-up only if the contact has prior engagement (stage >= cutoff) AND this stage's follow-up hasn't been sent yet (check deal_details.last_followup_stage). Load the template from template_path in the stage_actions entry (cat that policy document); if absent, use a short generic follow-up. After sending, record last_followup_stage in deal_details to prevent duplicates. Use email_send() tool to do it, and log the activity after (fetch log-crm-activity skill for details).
4. If inactivity >= archive_days (contact_last_outbound_ts < now - archive_days x 86400): set deal_lost_reason and move deal to Lost stage.
5. Summarize: how many emailed, closed as Lost, skipped — and why skipped. If any deal is in a stage that is at or after the cutoff but has no entry in stage_actions, print a widget to start a chat about it:

print_widget(type="start_chat", expert="default", text="Stage [stage_name] has stalled deals but no action defined — click to decide what to do.")