Kiro CLI MCP Server

Kiro CLI MCP Server

Enables IDE agents like Cursor and Windsurf to orchestrate kiro-cli with advanced session management, process pooling for 10x faster responses, and multi-project workflow support through isolated contexts.

Category
Visit Server

README

Kiro CLI MCP Server

A Model Context Protocol (MCP) server that enables IDE agents like Cursor and Windsurf to orchestrate kiro-cli with advanced session management, process pooling, and robust error handling.

Overview

Kiro CLI MCP Server bridges the gap between IDE agents and kiro-cli by providing a standardized MCP interface with enterprise-grade features:

  • 10x Performance Improvement: Process pooling reduces response time from ~500ms to ~50ms
  • Multi-Session Management: Isolated contexts for different projects/workflows
  • Production-Ready Reliability: Comprehensive error handling, timeout management, and process cleanup
  • Mock Mode: Development and testing without kiro-cli dependency

Features

Core Capabilities

  • Chat Integration: Send messages to kiro-cli and receive AI responses
  • Session Management: Create, switch, and manage multiple isolated sessions
  • Command Execution: Execute kiro-cli commands (/help, /mcp, etc.)
  • Custom Agents: Use and list available custom agents
  • History Management: Store and retrieve conversation history per session
  • Async Operations: Background task execution with progress polling

Performance & Reliability

  • Process Pooling: Reuse warm kiro-cli processes for 10x faster responses
  • Process Tree Cleanup: Prevent orphaned processes across platforms
  • Automatic Fallback: Mock mode when kiro-cli unavailable
  • Timeout Handling: Configurable timeouts with graceful cleanup
  • Session Isolation: Per-project working directories and conversation state

Installation

Prerequisites

  • Python 3.10+
  • kiro-cli installed and available in PATH (for full functionality - uses mock mode if unavailable)

From Source (Current Method)

git clone https://github.com/your-org/kiro-cli-mcp.git
cd kiro-cli-mcp
pip install -e .

Via pip (After PyPI Publication)

# Will be available after publishing to PyPI
pip install kiro-cli-mcp

Via uvx (After PyPI Publication)

# Will be available after publishing to PyPI
uvx install kiro-cli-mcp

Configuration

IDE Integration

Add to your IDE's MCP configuration file:

Cursor/Claude Desktop (~/.config/claude-desktop/mcp.json):

{
  "mcpServers": {
    "kiro-cli-mcp": {
      "command": "uvx",
      "args": ["kiro-cli-mcp"],
      "env": {
        "KIRO_MCP_LOG_LEVEL": "INFO"
      }
    }
  }
}

Windsurf (.windsurf/mcp.json):

{
  "mcpServers": {
    "kiro-cli-mcp": {
      "command": "python",
      "args": ["-m", "kiro_cli_mcp"],
      "env": {
        "KIRO_MCP_CLI_PATH": "/usr/local/bin/kiro-cli",
        "KIRO_MCP_POOL_SIZE": "5"
      },
      "autoApprove": [
        "kiro_session_list",
        "kiro_agents_list",
        "kiro_history"
      ]
    }
  }
}

Environment Variables

Variable Description Default
KIRO_MCP_CLI_PATH Path to kiro-cli executable kiro-cli
KIRO_MCP_COMMAND_TIMEOUT Command timeout (seconds) - IDE-optimized 30
KIRO_MCP_MAX_SESSIONS Maximum concurrent sessions 10
KIRO_MCP_SESSION_TIMEOUT Session idle timeout (seconds) 300
KIRO_MCP_CLEANUP_INTERVAL Session cleanup check interval (seconds) 30
KIRO_MCP_LOG_LEVEL Logging level INFO
KIRO_MCP_DEFAULT_MODEL Default AI model for kiro-cli claude-opus-4.5
KIRO_MCP_DEFAULT_AGENT Default agent to use kiro_default
KIRO_MCP_LOG_RESPONSE Log full CLI responses for debugging true
KIRO_MCP_POOL_SIZE Process pool size 5
KIRO_MCP_POOL_ENABLED Enable process pooling true
KIRO_MCP_POOL_IDLE_TIME Process idle time before recycling (seconds) 300
KIRO_MCP_POOL_MAX_USES Max uses per process before recycling 100
KIRO_MCP_MAX_ASYNC_TASKS Maximum concurrent async tasks 100
KIRO_MCP_TASK_TTL Task result TTL (seconds) 3600

Available MCP Tools

Session Management

  • kiro_session_create - Create new session with optional agent and working directory
  • kiro_session_list - List all active sessions
  • kiro_session_switch - Switch to specific session
  • kiro_session_end - End a session
  • kiro_session_clear - Clear session history files
  • kiro_session_save - Save session to file

Chat & Commands

  • kiro_chat - Send chat message and get AI response
  • kiro_command - Execute kiro-cli commands (/help, /mcp, etc.)
  • kiro_agents_list - List available custom agents

History Management

  • kiro_history - Get conversation history for session
  • kiro_history_clear - Clear conversation history

Async Operations

  • kiro_chat_async - Start background chat task
  • kiro_task_status - Poll task progress and results
  • kiro_task_cancel - Cancel running task
  • kiro_task_list - List active tasks

Monitoring

  • kiro_pool_stats - Get process pool performance statistics

Usage Examples

Basic Chat

# Create session for project
await mcp_client.call_tool("kiro_session_create", {
    "working_directory": "/path/to/project",
    "agent": "code-reviewer"
})

# Send message
response = await mcp_client.call_tool("kiro_chat", {
    "message": "Analyze this codebase and suggest improvements"
})

Multi-Project Workflow

# Project A
session_a = await mcp_client.call_tool("kiro_session_create", {
    "working_directory": "/projects/frontend",
    "agent": "react-expert"
})

# Project B  
session_b = await mcp_client.call_tool("kiro_session_create", {
    "working_directory": "/projects/backend", 
    "agent": "python-expert"
})

# Switch between projects
await mcp_client.call_tool("kiro_session_switch", {
    "session_id": session_a["session_id"]
})

Async Operations

# Start long-running task
task = await mcp_client.call_tool("kiro_chat_async", {
    "message": "Generate comprehensive test suite"
})

# Poll for progress
while True:
    status = await mcp_client.call_tool("kiro_task_status", {
        "task_id": task["task_id"]
    })
    if status["status"] == "completed":
        break
    await asyncio.sleep(1)

Architecture

MCP Protocol Integration

  • Server: Built on official MCP SDK (mcp.server.Server)
  • Transport: JSON-RPC 2.0 over stdio
  • Tools: 16 registered tools with schema validation
  • Resources: Minimal resource handling for extensibility

Process Management

IDE Agent → MCP Server → Process Pool → kiro-cli instances
                    ↓
              Session Manager → Isolated contexts per project

Key Components

  • SessionManager: Multi-session isolation and lifecycle management
  • ProcessPool: Warm process reuse for 10x performance improvement
  • CommandExecutor: Robust command execution with timeout handling
  • StreamingTaskManager: Async task execution with progress polling

Performance Optimizations

  1. Process Pooling: Reuse warm kiro-cli processes
  2. Session Affinity: Route requests to appropriate process
  3. Intelligent Cleanup: Remove idle/unhealthy processes
  4. Mock Mode: Fast responses during development

Development

Setup

git clone https://github.com/your-org/kiro-cli-mcp.git
cd kiro-cli-mcp
pip install -e ".[dev]"

Testing

# Run all tests
pytest

# With coverage
pytest --cov=kiro_cli_mcp --cov-report=html

# Property-based tests
pytest tests/test_config.py -v

Code Quality

# Format code
ruff format .

# Lint
ruff check .

# Type checking
mypy src/

Running Server

# Development mode with debug logging
python -m kiro_cli_mcp --log-level DEBUG

# With custom config
python -m kiro_cli_mcp --config config.json

Contributing

  1. Fork the repository
  2. Create feature branch (git checkout -b feature/amazing-feature)
  3. Commit changes (git commit -m 'Add amazing feature')
  4. Push to branch (git push origin feature/amazing-feature)
  5. Open Pull Request

Troubleshooting

kiro-cli Not Found

Server automatically enables mock mode if kiro-cli is unavailable:

# Check kiro-cli availability
which kiro-cli

# Set custom path
export KIRO_MCP_CLI_PATH=/custom/path/to/kiro-cli

# Verify server mode
python -m kiro_cli_mcp --log-level DEBUG
# Look for: "✅ kiro-cli is available" or "❌ kiro-cli not available: enabling mock mode"

Performance Issues

# Verify process pooling is enabled
python -m kiro_cli_mcp --log-level DEBUG
# Look for: "🔄 Using pooled process execution"

# Check pool statistics
# Use kiro_pool_stats tool to monitor performance

Session Management

# Increase session limits
export KIRO_MCP_MAX_SESSIONS=20
export KIRO_MCP_SESSION_TIMEOUT=7200  # 2 hours

# Clear stuck sessions
# Sessions auto-cleanup after timeout

Process Cleanup

If you encounter orphaned processes:

# Unix/Linux/macOS
pkill -f kiro-cli

# Windows  
taskkill /F /IM kiro-cli.exe

# Check process groups (Unix)
ps -eo pid,pgid,cmd | grep kiro

License

MIT License - see LICENSE file for details.

Support

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
E2B

E2B

Using MCP to run code via e2b.

Official
Featured