MCP Servers

 _ .-') _             .-')     _ (`-.    ('-.     .-') _             ('-. .-.,---. 
( (  OO) )           ( OO ).  ( (OO  )  ( OO ).-.(  OO) )           ( OO )  /|   | 
 \     .'_   ,-.-') (_)---\_)_.`     \  / . --. //     '._  .-----. ,--. ,--.|   | 
 ,`'--..._)  |  |OO)/    _ |(__...--''  | \-.  \ |'--...__)'  .--./ |  | |  ||   | 
 |  |  \  '  |  |  \\  :` `. |  /  | |.-'-'  |  |'--.  .--'|  |('-. |   .|  ||   | 
 |  |   ' |  |  |(_/ '..`''.)|  |_.' | \| |_.'  |   |  |  /_) |OO  )|       ||  .' 
 |  |   / : ,|  |_.'.-._)   \|  .___.'  |  .-.  |   |  |  ||  |`-'| |  .-.  |`--'  
 |  '--'  /(_|  |   \       /|  |       |  | |  |   |  | (_'  '--'\ |  | |  |.--.  
 `-------'   `--'    `-----' `--'       `--' `--'   `--'    `-----' `--' `--''--'  

What it is

Multi-agent MCP orchestrator for Cursor, Claude Code, or any MCP-compatible client. Describe a code review or audit task in plain English, Dispatch plans a DAG, runs specialist agents in parallel, critiques results, and returns one prioritized report.

How it works

You: "review this auth middleware for production readiness"

Dispatch:  plan → run agents (DAG) → critique → merge → report

Agents run in stages. Those with no dependencies run first; downstream agents get prior results as context.

Demo

npx dispatch-mcp

▶ dispatch  planning...
▶ dispatch  routing to [security] [architecture]
▶ dispatch  [security] done 3.1s
▶ dispatch  [architecture] done 3.4s
▶ dispatch  routing to [testing]
▶ dispatch  [testing] done 2.0s
▶ dispatch  critiquing...
▶ dispatch  merging...
✔ dispatch  done in 8.5s — 3 agents, 2 stages, 0 failures

Install

npx dispatch-mcp

Config

Add to .cursor/mcp.json (project root or ~/.cursor/mcp.json):

{
  "mcpServers": {
    "dispatch": {
      "command": "npx",
      "args": ["-y", "dispatch-mcp"],
      "env": {
        "ANTHROPIC_API_KEY": "your-api-key-here"
      }
    }
  }
}

Restart Cursor. Three tools: dispatch_run, dispatch_plan, dispatch_agent.

Adding agents

Open src/agents/registry.ts and add an entry. The Planner discovers it automatically.

makeAgent(
  "i18n",
  "Finds hardcoded strings that should be externalized for internationalization.",
  `You are an i18n specialist. Find hardcoded user-facing strings that should be 
moved to translation files. Format as a numbered list with file locations.`
),

Agents (9 built-in)

| Agent | Focus | |-------|-------| | security | Injection, auth, OWASP Top 10 | | performance | N+1, memory leaks, caching | | architecture | SOLID, coupling, design patterns | | testing | Coverage gaps, flaky tests | | documentation | Docstrings, naming, APIs | | accessibility | ARIA, WCAG 2.1 AA | | dependencies | CVEs, outdated, licenses | | i18n | Hardcoded strings, locale, RTL | | devops | CI/CD, Docker, secrets, observability |

More

• Env: ANTHROPIC_API_KEY required. Optional: DISPATCH_CACHE_DIR, DISPATCH_DISABLE_CACHE=1
• Local dev: .env with key, then npm run dev
• Examples: examples/ — sample inputs for each tool

MIT · GitHub