Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
aiskillstore avatar

Multi Agent Analysis

aiskillstore/marketplace

  • 1 installs
  • 404 repo stars
  • Updated August 5, 2026
  • aiskillstore/marketplace

multi-agent-analysis is a Claude Code skill that analyzes coordination models, handoffs, and state sharing in multi-agent systems.

About

multi-agent-analysis is a Claude Code skill that analyzes how multi-agent systems coordinate. A developer uses it to identify the coordination model, document how control transfers between agents, classify shared versus isolated state, and trace communication protocols. It includes Python examples of supervisor, peer-to-peer, pipeline, and market-based coordination.

  • Analyzes coordination patterns in multi-agent systems (supervisor, peer-to-peer, pipeline, market-based)
  • Documents handoff mechanisms and classifies state sharing (blackboard vs message passing)
  • Traces communication protocols and compares coordination models across frameworks

Multi Agent Analysis by the numbers

  • 1 all-time installs (skills.sh)
  • Ranked #14,102 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
At a glance

multi-agent-analysis capabilities & compatibility

Capabilities
multi agent analysis · coordination mapping · orchestration review
Use cases
orchestration · research
Pricing
Free
From the docs

What multi-agent-analysis says it does

Analyze coordination patterns, handoff mechanisms, and state sharing in multi-agent systems.
SKILL.md
1. **Identify coordination model** — Supervisor, peer-to-peer, pipeline
SKILL.md
npx skills add https://github.com/aiskillstore/marketplace --skill multi-agent-analysis

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1
repo stars★ 404
Last updatedAugust 5, 2026
Repositoryaiskillstore/marketplace ↗

What it does

Analyze coordination, handoffs, and state sharing in a multi-agent system.

Who is it for?

Developers evaluating or comparing multi-agent orchestration architectures.

Skip if: Building a single-agent tool or non-agent code analysis.

When should I use this skill?

You need to understand how agents transfer control, share state, or communicate in a multi-agent system.

What you get

A documented view of the coordination model, handoff mechanisms, and state-sharing pattern.

  • Coordination model classification
  • Handoff mechanism documentation
  • State-sharing analysis

By the numbers

  • 4 coordination models (supervisor, peer-to-peer, pipeline, market-based)
  • 3 handoff mechanisms (explicit, router-based, implicit state-based)

Files

SKILL.mdMarkdownGitHub ↗

Multi-Agent Analysis

Analyzes coordination patterns in multi-agent systems.

Process

1. Identify coordination model — Supervisor, peer-to-peer, pipeline 2. Document handoffs — How control transfers between agents 3. Classify state sharing — Blackboard vs message passing 4. Trace communication — Protocol and data flow

Coordination Models

Supervisor (Hierarchical)

        ┌─────────────┐
        │  Supervisor │
        │   (Router)  │
        └──────┬──────┘
               │
    ┌──────────┼──────────┐
    │          │          │
    ▼          ▼          ▼
┌───────┐  ┌───────┐  ┌───────┐
│Worker1│  │Worker2│  │Worker3│
│(Search)│  │(Code) │  │(Write)│
└───────┘  └───────┘  └───────┘
class Supervisor:
    def route(self, task: str) -> Agent:
        """Decide which worker handles the task"""
        if "search" in task:
            return self.search_agent
        elif "code" in task:
            return self.code_agent
        else:
            return self.general_agent
    
    def run(self, input: str):
        while not self.is_done():
            agent = self.route(self.current_task)
            result = agent.run(self.current_task)
            self.update_state(result)

Characteristics:

  • Central control point
  • Clear routing logic
  • Single point of failure
  • Easy to understand

Peer-to-Peer

┌───────┐     ┌───────┐
│Agent A│◄───►│Agent B│
└───┬───┘     └───┬───┘
    │             │
    │  ┌───────┐  │
    └─►│Agent C│◄─┘
       └───────┘
class PeerAgent:
    def __init__(self, peers: list["PeerAgent"]):
        self.peers = peers
    
    def delegate(self, task: str):
        """Find a peer that can handle this"""
        for peer in self.peers:
            if peer.can_handle(task):
                return peer.run(task)
        return self.run_locally(task)
    
    def broadcast(self, message: str):
        """Send to all peers"""
        for peer in self.peers:
            peer.receive(message)

Characteristics:

  • Decentralized
  • Resilient to single failures
  • Complex coordination
  • Harder to debug

Pipeline (Sequential)

┌───────┐    ┌───────┐    ┌───────┐    ┌───────┐
│Planner│───►│Executor│───►│Reviewer│───►│Output │
└───────┘    └───────┘    └───────┘    └───────┘
class Pipeline:
    def __init__(self, stages: list[Agent]):
        self.stages = stages
    
    def run(self, input):
        result = input
        for stage in self.stages:
            result = stage.run(result)
        return result

Characteristics:

  • Clear data flow
  • Easy to reason about
  • Limited parallelism
  • Each stage is a bottleneck

Market-Based

class MarketCoordinator:
    def __init__(self, agents: list[Agent]):
        self.agents = agents
    
    def auction(self, task: str):
        """Agents bid on tasks"""
        bids = []
        for agent in self.agents:
            bid = agent.bid(task)  # Returns confidence/cost
            bids.append((agent, bid))
        
        # Select winner
        winner = max(bids, key=lambda x: x[1])
        return winner[0].run(task)

Characteristics:

  • Dynamic allocation
  • Self-organizing
  • Overhead of bidding
  • Complex to tune

Handoff Mechanisms

Explicit Transfer

class Agent:
    def handoff_to(self, target: "Agent", context: dict):
        """Explicit control transfer"""
        return HandoffResult(
            target_agent=target,
            context=context,
            return_control=True
        )
    
    def run(self, input):
        result = self.think(input)
        if result.needs_specialist:
            return self.handoff_to(
                self.get_specialist(result.domain),
                context={"original_task": input, "progress": result}
            )
        return result

Router-Based

class Router:
    def __init__(self, agents: dict[str, Agent]):
        self.agents = agents
        self.routing_llm = LLM()
    
    def route(self, input: str) -> Agent:
        decision = self.routing_llm.generate(f"""
        Given this input: {input}
        Which agent should handle it?
        Options: {list(self.agents.keys())}
        """)
        return self.agents[decision.agent_name]

Implicit (State-Based)

class StateBasedCoordinator:
    def run(self, input):
        state = {"input": input, "stage": "planning"}
        
        while state["stage"] != "done":
            # Agent selection based on state
            agent = self.get_agent_for_stage(state["stage"])
            result = agent.run(state)
            state = self.update_state(state, result)
        
        return state["output"]

State Sharing Patterns

Blackboard (Shared Global State)

class Blackboard:
    """Shared state all agents can read/write"""
    def __init__(self):
        self.state = {}
        self.lock = threading.Lock()
    
    def read(self, key: str):
        return self.state.get(key)
    
    def write(self, key: str, value):
        with self.lock:
            self.state[key] = value

# Agents share the blackboard
blackboard = Blackboard()
agent_a = Agent(blackboard)
agent_b = Agent(blackboard)

Pros: Simple, full visibility Cons: Race conditions, tight coupling, hard to scale

Message Passing (Isolated State)

class Agent:
    def __init__(self):
        self.inbox = Queue()
        self.state = {}  # Private state
    
    def send(self, target: "Agent", message: dict):
        target.inbox.put(message)
    
    def receive(self) -> dict:
        return self.inbox.get()
    
    def run(self):
        while True:
            message = self.receive()
            result = self.process(message)
            if message.get("reply_to"):
                self.send(message["reply_to"], result)

Pros: Isolation, clear boundaries, scalable Cons: More complex, async handling

Hybrid

class HybridCoordinator:
    def __init__(self, agents):
        # Shared read-only context
        self.shared_context = {"tools": [...], "config": {...}}
        
        # Per-agent mutable state
        self.agent_states = {a.id: {} for a in agents}
        
        # Message queues for communication
        self.queues = {a.id: Queue() for a in agents}

Communication Protocol Analysis

Direct Invocation

result = agent_b.run(input)

Latency: Lowest Coupling: Highest Async: No

Queue-Based

task_queue.put(task)
# ... later ...
result = result_queue.get()

Latency: Medium Coupling: Low Async: Yes

Event-Driven

event_bus.emit("task:created", task)

@event_bus.on("task:created")
def handle_task(task):
    result = process(task)
    event_bus.emit("task:completed", result)

Latency: Variable Coupling: Lowest Async: Yes

Output Template

## Multi-Agent Analysis: [Framework Name]

### Coordination Model
- **Type**: [Supervisor/Peer-to-Peer/Pipeline/Market]
- **Central Control**: [Yes/No]
- **Location**: `path/to/orchestrator.py`

### Agent Inventory

| Agent | Role | Can Delegate To |
|-------|------|-----------------|
| Supervisor | Routing | All workers |
| SearchAgent | Web search | None |
| CodeAgent | Code execution | Reviewer |

### Handoff Mechanism
- **Type**: [Explicit/Router/Implicit]
- **Bidirectional**: [Yes/No]
- **Context Preserved**: [Full/Partial/Minimal]

### State Sharing
- **Pattern**: [Blackboard/Message/Hybrid]
- **Shared State**: [List what's shared]
- **Isolation Level**: [None/Partial/Full]

### Communication Protocol
- **Method**: [Direct/Queue/Event]
- **Async**: [Yes/No]
- **Location**: `path/to/comms.py`

### Loop Prevention
- **Mechanism**: [Depth limit/Visited set/None]
- **Max Handoffs**: [N or Unlimited]

Integration

  • Prerequisite: codebase-mapping to identify agent files
  • Feeds into: comparative-matrix for coordination decisions
  • Related: control-loop-extraction for individual agent loops

Related skills

mattpocockSetup Matt Pocock SkillsScaffold the per-repo configuration that Matt Pocock’s engineering agent skills rely on so they understand the issue tracker, triage labels, and domain documentation la495k196kopen.feishu.cnLark AppsConnect Claude, Cursor or custom agents directly to Lark (Feishu) for messaging, document automation, approval workflows and enterprise data access.402kjuliusbrusseeCavemanSlash token usage by roughly 75% while keeping every technical detail intact when working with Claude Code, Cursor or similar agents.391k94.7klarksuiteLark Skill MakerQuickly turn any Lark/Feishu OpenAPI call or multi-step workflow into a reusable agent skill with its own SKILL.md.386k16kxixu-meRunning Claude Code Via Litellm CopilotRun Claude Code at a fraction of the cost by routing requests through LiteLLM to the GitHub Copilot Chat API.270k72runcomfy-comCodex PetGenerate a complete Codex Pet spritesheet and metadata from one reference image without needing an OpenAI key or Codex Pro.246k10

FAQ

Which coordination models does it cover?

Supervisor (hierarchical), peer-to-peer, pipeline (sequential), and market-based coordination, each with characteristics and Python examples.

How does it classify state sharing?

It distinguishes blackboard-style shared global state from message-passing, and covers handoff mechanisms like explicit transfer, router-based, and implicit state-based.

AI & Agent Buildingagentsresearch
Browse by category
AI & AgentsFrontend & DesignBackend & DataDev ToolsAutomationTesting & ReviewSecurityMonitoring & CloudDocs & Planning
Design & frontend
Design systemsAnimation & motionAccessibility (a11y)CSS & TailwindUI designUXDesign to codeshadcn/uiLanding pagesBest design skills
Most searched skills
find-skillscavemanfrontend-designbrainstormingsystematic-debuggingstop-slopemil-design-engui-ux-pro-maxkarpathy-guidelinesweb-design-guidelinesreact-best-practicesgrill-meimprovemicrosoft-foundryagent-browser
Browse all listings
All agent skillsAll MCP serversAll marketplacesAll plugins

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.