amcos-emergency-handoff

name	amcos-emergency-handoff
description	Use when handling emergency work handoff from failed agents. Trigger with agent failure handoff or task blocker routing. Loaded by ai-maestro-chief-of-staff-main-agent
user-invocable	false
license	Apache-2.0
compatibility	Requires AI Maestro installed.
metadata	{"author":"Emasoft","version":"1.0.0"}
context	fork
agent	ai-maestro-chief-of-staff-main-agent

AMCOS Emergency Handoff

Overview

Transfer critical work when deadlines cannot wait. Routes task blocker escalations to AMAMA.

Prerequisites

• Deadline-critical failure OR task blocker escalation from AMOA
• AI Maestro running locally
• Handoff dir: $CLAUDE_PROJECT_DIR/thoughts/shared/handoffs/emergency/

Instructions

Phase 5: Emergency Work Handoff

1. Identify critical tasks at risk
2. Notify AMOA of emergency handoff initiation
3. Find an available receiving agent
4. Create minimum-viable handoff doc (task, state, files, deadline)
5. Send handoff via AI Maestro
6. Verify receiving agent acknowledges
7. Notify AMAMA (notification only, no approval needed)
8. Monitor deadline compliance

Aspect	Regular	Emergency
Timing	After replacement	Immediately
Completeness	Full context	Minimum viable
Recipient	Replacement	Any available
Duration	Permanent	Temporary

Task Blockers vs Agent Failures

Agent Failures -- resolve via failure recovery workflow.

Task Blockers -- blocked by missing info/access/user decision:

1. Receive escalation from AMOA
2. AMCOS can resolve? (reassignment, permission) -> handle directly
3. Needs user input? -> route to AMAMA via blocker-escalation
4. Track; route AMAMA response back to AMOA

Decision Tree

Escalation -> Agent failure? -> recovery workflow
           -> Blocker AMCOS resolves? -> handle
           -> Needs user input? -> route to AMAMA

Handoff Validation Checklist

Copy this checklist and track your progress:

• Required fields: from/to/type/UUID/task/failed_agent/failure_reason
• UUID unique, target alive, files exist
• No [TBD] placeholders, deadline and acceptance criteria set

Output

Result	Output
Handoff sent	Work transferred to available agent
Blocker routed	Awaiting AMAMA user decision
Blocker resolved	Agent reassigned directly

Error Handling

Issue	Resolution
No available agent	Notify AMAMA for manual intervention
Handoff rejected	Try next available agent
Deadline missed	Document, notify, post-mortem
AMAMA unresponsive	15 min wait, reminder, escalate

Examples

Emergency handoff

amp-send.sh apps-svgplayer-development "EMERGENCY: SVG fix (2h)" urgent \
  '{"type":"emergency-handoff","message":"svgbbox crashed. Fix by 18:00 UTC.","task":"Fix SVG rendering","failed_agent":"libs-svg-svgbbox","deadline":"2025-02-05T18:00:00Z"}'

Task blocker routing

amp-send.sh amama-assistant-manager "BLOCKER: API key needed" high \
  '{"type":"blocker-escalation","message":"API key needed, user must provide.","blocker_type":"user-decision","impact":"Deploy blocked","escalated_from":"amoa-deployment"}'

Checklist

Copy this checklist and track your progress:

• Identify critical tasks at risk from failure
• Find available receiving agent and create handoff doc
• Send handoff and verify acknowledgment
• Notify AMAMA and monitor deadline compliance

Resources

• work-handoff-during-failure — Emergency handoff procedures, message formats, reconciliation
  • 5.1 When to use this document
  • 5.2 Overview of emergency handoff
  • 5.3 Triggering emergency handoff
    • 5.3.1 When to initiate emergency handoff
    • 5.3.2 Notification to orchestrator
    • 5.3.3 Notification to manager
  • 5.4 Creating emergency handoff documentation
    • 5.4.1 Required handoff content
    • 5.4.2 Handoff document template
    • 5.4.3 Extracting information from failed agent
  • 5.5 Reassigning work during failure
    • 5.5.1 Temporary reassignment to other agents
    • 5.5.2 Holding tasks for replacement agent
    • 5.5.3 Splitting tasks across multiple agents
  • 5.6 Emergency handoff message formats
    • 5.6.1 Handoff request to orchestrator
    • 5.6.2 Task reassignment notification
    • 5.6.3 Receiving agent instructions
  • 5.7 Post-failure work reconciliation
    • 5.7.1 Identifying duplicate work
    • 5.7.2 Merging partial progress
    • 5.7.3 Updating task status