multi-agent-architecture-reference

v2026.09.24

Decision matrix for selecting multi-agent topologies (Supervisor, Swarm, Hierarchical, Conductor) with token economics, failure modes, and escalation paths

GitHub
安装命令
npx skhub add oimiragieo/multi-agent-architecture-reference
Markdown
SKILL.md

Multi-Agent Architecture Reference

<identity> Canonical reference for multi-agent topology selection — provides a 6-topology decision matrix with token economics, failure modes, escalation paths, and links to existing agent-studio patterns. </identity> <capabilities> - Select the optimal multi-agent topology for a given task based on complexity, cost constraints, and fault tolerance requirements - Estimate token cost multiplier for each topology relative to single-agent baseline - Identify known failure modes (SE-M01 through SE-M05) and their mitigations - Map tasks to existing agent-studio patterns (wave-executor, consensus-voting, swarm-coordination) - Provide escalation path guidance: when to upgrade TRIVIAL → Supervisor → Hierarchical - Reference conductor pattern as agent-studio's default recommendation </capabilities> <instructions>

Step 1: Characterize the Task

Answer these four questions before selecting a topology:

  1. Task independence: Can sub-tasks run in parallel without shared state? (YES → Swarm or Fan-out)
  2. Task types known: Is the set of task types stable and deterministic at design time? (YES → Supervisor)
  3. Phase complexity: Does the work require multi-stage sub-orchestration? (YES → Hierarchical or Conductor)
  4. Stakes: Does an incorrect outcome require multi-reviewer agreement? (YES → Consensus Voting)

Step 2: Apply the Topology Decision Matrix

TopologyToken CostBest ForFailure ModesExisting Skill
Conductor~6xSequential phases, ordered agent steps, default agent-studio patternOrchestrator overload (SE-M01)master-orchestrator.md
Supervisor~5xKnown task types, specialist agents, deterministic routingSingle point of failure; router miscalibration (SE-M01)Built into Router
Fan-out/Fan-in~8xParallel review/analysis, map-reduce, searchResult aggregation complexitywave-executor
Swarm~8xIndependent tasks, load balancing, fault-tolerant processingCoordination overhead; consensus deadlock; orphaned tasks (SE-M02, SE-M05)swarm-coordination
Consensus Voting~12xHigh-stakes decisions requiring multi-reviewer agreementDeadlock on split votes (SE-M02)consensus-voting
Hierarchical~15xEPIC complexity, multiple distinct phases with sub-orchestrationCascade failures; token runaway at depth >3 (SE-M03, SE-M04)Custom per project

Token costs are relative to single-agent baseline (as of 2026). Use as order-of-magnitude guidance.

Step 3: Check Failure Mode Taxonomy

Before finalizing topology, verify mitigation for relevant failure modes:

SE-M01: Coordinator Overload

  • Topologies affected: Supervisor, Conductor, Hierarchical root
  • Symptom: Single coordinator receives more traffic than it can route
  • Fix: Distribute coordination or add routing replicas; use wave-executor for fan-out

SE-M02: Swarm Deadlock

  • Topologies affected: Swarm, Consensus Voting
  • Symptom: Agents wait for each other's consensus indefinitely
  • Fix: Timeout + majority-vote with tie-breaker; set consensus_timeout_ms

SE-M03: Cascade Failure

  • Topologies affected: Hierarchical
  • Symptom: A mid-level agent failure halts all downstream agents
  • Fix: Circuit breakers at each tier; retry with backoff; fallback agents

SE-M04: Token Runaway

  • Topologies affected: Hierarchical
  • Symptom: Spawning too many levels burns tokens exponentially
  • Fix: Set max_depth=3; monitor token budget per level; prefer Conductor over deep Hierarchical

SE-M05: Orphaned Tasks

  • Topologies affected: Swarm
  • Symptom: Agents drop tasks when no ownership is clear
  • Fix: Assign task IDs; use TaskUpdate tracking; require TaskUpdate(in_progress) on pickup

Step 4: Apply Escalation Path

Use the complexity escalation ladder when initial topology is insufficient:

TRIVIAL → Single agent (no multi-agent needed)
    ↓ (task types > 1, > 3 files)
LOW → Supervisor (router delegates to 2-3 specialists)
    ↓ (parallel processing needed)
MEDIUM → Conductor + Fan-out (master-orchestrator + wave-executor)
    ↓ (multi-phase with sub-orchestration)
HIGH → Hierarchical (orchestrators at multiple tiers)
    ↓ (high-stakes decision required)
EPIC → Hierarchical + Consensus Voting (max 3 tiers + voting gate)

Step 5: Reference Existing agent-studio Patterns

PatternSkill/FileUse Case
Conductor (DEFAULT).claude/agents/orchestrators/master-orchestrator.mdSequential phase execution; TaskUpdate coordination
Fan-out/Fan-inwave-executor skillParallel batch processing; EPIC-tier pipelines
Swarmswarm-coordination skillConcurrent independent task execution
Consensusconsensus-voting skillHigh-stakes decisions; multi-reviewer agreement
SupervisorBuilt into CLAUDE.mdTask routing to specialist agents

When in doubt, start with Conductor. The master-orchestrator pattern drives sequential phases with explicit TaskUpdate coordination — the lowest-risk default for most MEDIUM/HIGH tasks.

</instructions> <examples>

Example 1: Code Review Pipeline

  • Task: Review 5 files for security, quality, and style
  • Character: Tasks are independent (YES), parallel OK (YES)
  • Topology: Fan-out/Fan-in (~8x)
  • Pattern: wave-executor skill — spawn 3 reviewers in parallel, aggregate results

Example 2: Feature Implementation

  • Task: Design → Implement → Test → Document
  • Character: Sequential phases, ordered steps (YES)
  • Topology: Conductor (~6x)
  • Pattern: master-orchestrator with TaskUpdate coordination between phases

Example 3: Architecture Decision

  • Task: Choose between 3 database options for production system
  • Character: High stakes, requires agreement (YES)
  • Topology: Consensus Voting (~12x)
  • Pattern: consensus-voting skill — 3 architect agents vote, majority decides

Example 4: Batch Agent Creation

  • Task: Create 10 new agents from specs
  • Character: Independent tasks (YES), fault tolerance > ordering (YES)
  • Topology: Swarm (~8x)
  • Pattern: swarm-coordination skill with task ID assignment per agent
</examples>

<best_practices>

  • Default to Conductor (master-orchestrator) — it is the lowest-risk pattern for most tasks
  • Never use Hierarchical beyond depth=3 (token runaway risk SE-M04)
  • Always assign TaskUpdate(in_progress) on task pickup in Swarm to prevent SE-M05
  • Use Fan-out (wave-executor) instead of Swarm when tasks have clear aggregation boundary
  • Add consensus gate only for genuinely high-stakes decisions — 12x token cost is significant
  • Document token budget per topology tier when spawning Hierarchical
  • Cross-reference failure mode taxonomy before finalizing topology choice </best_practices>

Iron Laws

  1. ALWAYS start with Conductor — default to master-orchestrator for MEDIUM/HIGH tasks; only escalate to Hierarchical when sub-orchestration is explicitly required by the task structure.
  2. NEVER exceed depth=3 in Hierarchical — token cost grows exponentially at each tier; depth >3 triggers SE-M04 (token runaway) and is considered an architectural defect.
  3. ALWAYS assign TaskUpdate(in_progress) on Swarm task pickup — missing task ownership is the root cause of SE-M05 (orphaned tasks); every agent in a swarm must call TaskUpdate before doing work.
  4. NEVER use Consensus Voting for low-stakes decisions — 12x token multiplier is justified only for architecture decisions, security approvals, or irreversible production changes.
  5. ALWAYS cross-reference the failure mode taxonomy before finalizing topology — each topology has documented failure modes (SE-M01 through SE-M05); skipping this review leads to production incidents.

Anti-Patterns

Anti-PatternProblemFix
Defaulting to Hierarchical for every complex taskToken runaway at depth >3; cascade failure risk; over-engineering most tasksUse Conductor (sequential phases) first; only escalate to Hierarchical when sub-orchestration is mandatory
Using Swarm for ordered, dependent tasksSwarm agents run concurrently and cannot enforce ordering; produces race conditionsUse Conductor or Fan-out/Fan-in when task ordering matters
Skipping TaskUpdate(in_progress) in SwarmTasks become orphaned (SE-M05); no ownership tracking; duplicated or dropped workRequire every swarm agent to call TaskUpdate(in_progress) as its first action
Adding Consensus Voting speculatively12x token overhead kills budget for non-critical decisions; slowdown on all downstream tasksReserve consensus gate for genuinely high-stakes, irreversible decisions only
Mixing topology concerns (Supervisor + Swarm + Hierarchical in one flow)Complexity explosion; routing ambiguity; impossible to debug failuresPick one primary topology per orchestration scope; compose only at well-defined phase boundaries

Memory Protocol (MANDATORY)

Before starting:

Read .claude/context/memory/learnings.md to check for prior multi-agent architecture decisions.

After completing:

  • New topology decision → Append to .claude/context/memory/decisions.md
  • Failure mode encountered → Append to .claude/context/memory/issues.md
  • New pattern discovered → Append to .claude/context/memory/learnings.md

ASSUME INTERRUPTION: Your context may reset. If it's not in memory, it didn't happen.

Related Skills

  • wave-executor — Fan-out/Fan-in implementation
  • swarm-coordination — Swarm topology execution
  • consensus-voting — Byzantine consensus for high-stakes decisions
  • architecture-review — Validate topology choices against NFRs
  • complexity-assessment — Determine complexity level before topology selection
发现
标签

此技能尚未发布标签。

版本
最新版本元数据

版本

v2026.09.24

发布时间

2026年9月24日

分类

未分类

许可证

未指定

源路径

.claude/skills/multi-agent-architecture-reference

默认分支

main

最新提交

64b580e

Tree SHA

42a1df4