agentkit-mesh

Agent-to-agent discovery and delegation via MCP (Model Context Protocol).

Agents register their capabilities, discover each other by semantic search, and delegate tasks — all through standard MCP tools.

Quick Start

npx agentkit-mesh

This starts an MCP server over stdio, ready to connect to Claude Desktop, OpenClaw, or any MCP client.

MCP Configuration

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "agentkit-mesh": {
      "command": "npx",
      "args": ["agentkit-mesh"]
    }
  }
}

OpenClaw

Add to your OpenClaw config:

mcp:
  agentkit-mesh:
    command: npx agentkit-mesh

Architecture

┌─────────────┐     MCP      ┌──────────────────┐
│  AI Agent A  │◄────────────►│                  │
└─────────────┘              │  agentkit-mesh   │
                             │                  │
┌─────────────┐     MCP      │  ┌────────────┐  │
│  AI Agent B  │◄────────────►│  │  Registry   │  │
└─────────────┘              │  │  (SQLite)   │  │
                             │  └────────────┘  │
┌─────────────┐     MCP      │  ┌────────────┐  │
│  AI Agent C  │◄────────────►│  │  Discovery  │  │
└─────────────┘              │  └────────────┘  │
                             │  ┌────────────┐  │
                             │  │ Delegation  │  │
                             │  └────────────┘  │
                             └──────────────────┘

MCP Tools

mesh_register

Register an agent with its capabilities.

Parameter	Type	Description
name	string	Unique agent name
description	string	What this agent does
capabilities	string[]	List of capabilities
endpoint	string	Agent's MCP endpoint URL

mesh_discover

Discover agents matching a natural language query.

Parameter	Type	Description
query	string	Search query (e.g. "budget management")
limit	number?	Max results to return

Returns agents ranked by relevance score with matched capability terms.

mesh_unregister

Remove an agent from the registry.

Parameter	Type	Description
name	string	Agent name to remove

mesh_delegate

Delegate a task to another agent by name.

Parameter	Type	Description
targetName	string	Name of the target agent
task	string	Task description to delegate
context	string?	Optional JSON context

Connects to the target agent's MCP endpoint and calls its handle_task tool.

Use Case: FormBridge

An HR agent filling an expense form discovers the Finance agent:

import { AgentRegistry, DiscoveryEngine } from 'agentkit-mesh';

const registry = new AgentRegistry();

// Agents register themselves
registry.register({
  name: 'finance-agent',
  description: 'Budget management and expense approval',
  capabilities: ['budget', 'cost_center', 'expense_approval'],
  endpoint: 'http://localhost:4002/mcp',
});

// HR agent discovers who can help with budget fields
const discovery = new DiscoveryEngine();
const results = discovery.discover('budget cost center', registry);
// → [{ agent: finance-agent, score: 0.67, matchedTerms: ['budget', 'cost', 'center'] }]

See examples/ for a runnable demo.

Optional: Lore Integration

For semantic search beyond keyword matching, connect to a Lore server:

import { LoreDiscoveryEngine } from 'agentkit-mesh';

const engine = new LoreDiscoveryEngine('http://lore:8080', registry, 'api-key');
const results = await engine.discover('financial planning');
// Falls back to text matching if Lore is unavailable

Programmatic API

import { AgentRegistry, DiscoveryEngine, DelegationClient, createServer } from 'agentkit-mesh';

All classes are exported for direct use without the MCP server layer.

🧰 AgentKit Ecosystem

Project	Description
AgentLens	Observability & audit trail for AI agents
Lore	Cross-agent memory and lesson sharing
AgentGate	Human-in-the-loop approval gateway
FormBridge	Agent-human mixed-mode forms
AgentEval	Testing & evaluation framework
agentkit-mesh	Agent discovery & delegation	⬅️ you are here
agentkit-cli	Unified CLI orchestrator
agentkit-guardrails	Reactive policy guardrails

License

ISC