Table of Contents

• Overview
• When to Use
• Error Categories
• Escalation Ladder
• Quick Reference
• Module Reference
• Integration Pattern
• Exit Criteria

Damage Control

Overview

Adapts service-level error patterns from leyline:error-patterns to agent-level recovery in multi-agent coordination scenarios. Provides structured playbooks for the four most common agent failure modes.

When To Use

• Multi-agent coordination encounters agent failures
• Context windows overflow during complex tasks
• Merge conflicts arise from parallel agent work
• Partial failures need triage and salvage
• Orchestrators need standardized recovery procedures

When NOT To Use

• Single-agent work (use leyline:error-patterns directly)
• Service-level errors without agent coordination (API failures, rate limits)
• Simple retry scenarios handled by existing error patterns

Error Categories

Agent-level error categories extend the service-level taxonomy from leyline:error-patterns:

Agent Category	Service Equivalent	Severity	Recovery
AGENT_CRASH	PERMANENT	Critical	Replace agent, reassign tasks
CONTEXT_OVERFLOW	RESOURCE	Error	Graceful handoff, context shed
MERGE_CONFLICT	TRANSIENT	Warning	Stop, resolve, resume
PARTIAL_FAILURE	Mixed	Varies	Triage by sub-category

Escalation Ladder

Three-tier escalation with clear handoff points:

Tier 1: Agent Self-Recovery
  Agent detects issue, attempts automated recovery
  (retry, context shed, conflict resolution)
        |
        | Fails after 2 attempts
        v
Tier 2: Lead Agent Intervention
  Lead reassigns tasks, spawns replacement agents,
  coordinates resolution across team
        |
        | Cannot resolve or CRITICAL severity
        v
Tier 3: Human Escalation
  Human notified with full context, recommended actions,
  and current state summary

Quick Reference

Scenario	Module	First Action
Agent stops responding	agent-crash-recovery	Check heartbeat, reassign tasks
Response truncation	context-overflow	Summarize state, create continuation
File conflicts after parallel work	merge-conflict-resolution	Stop agents, lead resolves
Some tasks fail, others succeed	partial-failure-handling	Triage by error category

Module Reference

• agent-crash-recovery.md: Orphaned task detection, "replace don't wait" doctrine, state recovery
• context-overflow.md: Detection signals, graceful handoff, progressive context shedding
• merge-conflict-resolution.md: Post-execution conflict detection, resolution protocol, prevention
• partial-failure-handling.md: Error triage, per-category recovery, salvage protocol

Integration Pattern

# In your skill's frontmatter
dependencies: [leyline:damage-control]

Consumers should reference specific modules for targeted recovery:

# In orchestrator or coordination skill
On agent crash: follow `leyline:damage-control/modules/agent-crash-recovery.md`
On context overflow: follow `leyline:damage-control/modules/context-overflow.md`

Exit Criteria

• Failed agent identified and replaced (or escalated)
• Orphaned tasks reassigned to healthy agents
• Successful work preserved and committed
• Recovery actions logged for post-mortem