Claude Thread Continuity

Claude Thread Continuity

Automatically saves and restores project state when Claude threads hit token limits, ensuring seamless conversation continuity and preventing project fragmentation with intelligent name validation.

Category
访问服务器

README

🧠 Claude Thread Continuity MCP Server

Never lose context again! This MCP server automatically saves and restores project state when Claude threads hit token limits, ensuring seamless conversation continuity.

🚀 Features

  • 🔄 Automatic State Persistence - Auto-saves project context during conversations
  • ⚡ Seamless Restoration - Instantly restore full context when starting new threads
  • 🛡️ Smart Validation - Prevents project fragmentation with intelligent name checking
  • 🔒 Privacy First - All data stored locally on your machine
  • 🎯 Zero Configuration - Works invisibly once set up
  • 📊 Smart Triggers - Auto-saves on file changes, decisions, milestones
  • 🗂️ Multi-Project Support - Manage multiple concurrent projects

✨ NEW: Anti-Fragmentation System

Version 1.1 introduces intelligent project validation to prevent the common issue of accidentally creating multiple similar projects:

  • 🔍 Fuzzy Name Matching - Detects similar project names (70% similarity threshold)
  • ⚠️ Validation Warnings - Suggests consolidation when similar projects exist
  • 💪 Force Override - Bypass validation when genuinely different projects needed
  • 🎯 Configurable Thresholds - Adjust sensitivity for your workflow

Example Validation in Action

❌ Project "Hebrew Speaking Evaluation MVP" blocked
✅ Similar project found: "Hebrew Evaluation MVP" (85% similar)
🎯 Recommendation: Update existing project or use force=true

⚡ Quick Start

# 1. Clone the repository
git clone https://github.com/peless/claude-thread-continuity.git
cd claude-thread-continuity

# 2. Install dependencies
pip install -r requirements.txt

# 3. Test the enhanced server
python3 test_server.py

# 4. Add to Claude Desktop config
# See setup instructions below

🛠️ Installation

1. Install the MCP Server

# Create permanent directory
mkdir -p ~/.mcp-servers/claude-continuity
cd ~/.mcp-servers/claude-continuity

# Copy files (or clone repo to this location)
# Place server.py and requirements.txt here

2. Configure Claude Desktop

Edit your Claude Desktop configuration file:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\\Claude\\claude_desktop_config.json
  • Linux: ~/.config/Claude/claude_desktop_config.json

Add this configuration:

{
  "mcpServers": {
    "claude-continuity": {
      "command": "python3",
      "args": ["~/.mcp-servers/claude-continuity/server.py"],
      "env": {}
    }
  }
}

3. Restart Claude Desktop

Close and reopen Claude Desktop. The continuity tools will now be available automatically.

🎯 How It Works

Automatic Context Saving

The server automatically saves project state when:

  • ✅ Files are created or modified
  • ✅ Technical decisions are made
  • ✅ Project milestones are reached
  • ✅ Every 10 messages (fallback)

Smart Validation Process

Before saving, the system:

  1. Checks for Similar Names - Uses fuzzy matching to find existing projects
  2. Calculates Similarity - Compares project names with 70% threshold
  3. Provides Recommendations - Suggests consolidation or renaming
  4. Allows Override - Use force: true for edge cases

Context Restoration

When starting a new thread:

  1. Load Project: load_project_state: project_name="your-project"
  2. Full Context Restored: All technical decisions, files, and progress restored
  3. Continue Seamlessly: Pick up exactly where you left off

🔧 Available Commands

Command Description NEW in v1.1
save_project_state Save current project state ✨ Now with validation
load_project_state Restore full project context
list_active_projects View all tracked projects
get_project_summary Get quick project overview
validate_project_name Check for similar project names ✨ NEW
auto_save_checkpoint Triggered automatically

💡 Usage Examples

Starting a New Project (with Validation)

save_project_state: project_name="my-web-app", current_focus="Setting up React components", technical_decisions=["Using TypeScript", "Vite for bundling"], next_actions=["Create header component", "Set up routing"]

Checking Name Before Creating

validate_project_name: project_name="my-webapp", similarity_threshold=0.7

Force Override When Needed

save_project_state: project_name="my-web-app-v2", force=true, current_focus="Starting version 2"

Continuing After Token Limit

load_project_state: project_name="my-web-app"

Viewing All Projects

list_active_projects

🗂️ Data Storage

Project states are stored locally at:

~/.claude_states/
├── project-name-1/
│   ├── current_state.json
│   └── backup_*.json
└── project-name-2/
    ├── current_state.json
    └── backup_*.json
  • Privacy: Everything stays on your machine
  • Backups: Automatic backup rotation (keeps last 5)
  • Format: Human-readable JSON files
  • Validation: Metadata tracks validation bypass status

🏗️ Project State Structure

Each saved state includes:

{
  "project_name": "my-project",
  "current_focus": "What you're working on now",
  "technical_decisions": ["Key choices made"],
  "files_modified": ["List of files created/changed"],
  "next_actions": ["Planned next steps"],
  "conversation_summary": "Brief context summary",
  "last_updated": "2025-06-15T10:30:00Z",
  "version": "1.1",
  "validation_bypassed": false
}

🛡️ Validation Configuration

Default Settings

  • Similarity Threshold: 70% (0.7)
  • Comparison Method: Fuzzy string matching
  • Auto-save Behavior: Bypasses validation (uses force=true)

Customizing Validation

validate_project_name: project_name="test-project", similarity_threshold=0.8

Higher threshold = stricter matching (0.9 = 90% similar required) Lower threshold = looser matching (0.5 = 50% similar triggers warning)

🔍 Troubleshooting

Tools Not Appearing

  1. Check Claude Desktop logs
  2. Verify Python 3 is in your PATH: python3 --version
  3. Validate JSON config syntax
  4. Restart Claude Desktop completely

Testing the Enhanced Server

cd ~/.mcp-servers/claude-continuity
python3 test_server.py

The test suite now includes validation testing and will report:

  • ✅ Basic functionality tests
  • ✅ Project validation tests
  • ✅ Fuzzy matching accuracy
  • ✅ Force override functionality

Common Issues

Validation Too Strict: Lower the similarity threshold or use force=true

Permission Errors:

chmod +x ~/.mcp-servers/claude-continuity/server.py

Python Path Issues: Update the config to use full Python path:

{
  "command": "/usr/bin/python3",
  "args": ["~/.mcp-servers/claude-continuity/server.py"]
}

🧪 Development

Requirements

  • Python 3.8+
  • MCP SDK 1.0+
  • difflib (built-in, for fuzzy matching)

Running Tests

python3 test_server.py

Enhanced test suite includes:

  • Basic functionality validation
  • NEW: Project name similarity testing
  • NEW: Validation workflow testing
  • NEW: Force override testing
  • NEW: MCP tool validation

Project Structure

claude-thread-continuity/
├── server.py           # Main MCP server (enhanced with validation)
├── requirements.txt    # Python dependencies
├── test_server.py     # Comprehensive test suite
├── README.md          # This file
├── LICENSE            # MIT License
└── examples/          # Usage examples

🤝 Contributing

Contributions welcome! Please:

  1. Fork the repository
  2. Create a feature branch
  3. Add tests for new functionality
  4. Submit a pull request

Current Development Priorities

  • [ ] Integration with external project management tools
  • [ ] Advanced similarity algorithms
  • [ ] Project merging utilities
  • [ ] Custom validation rules

📄 License

MIT License - see LICENSE file for details.

🚀 Why This Matters

Before v1.1: 😫 Hit token limit → Lose all context → Re-explain everything → Lose momentum

Common Problem: 😤 Create "Hebrew MVP", then "Hebrew Evaluation MVP", then "Hebrew Speaking MVP" → Context scattered across multiple projects

After v1.1: 😎 Hit token limit → Start new thread → load_project_state → Continue seamlessly + Smart validation prevents fragmentation

Perfect for:

  • 🏗️ Complex Development Projects - Keep track of architecture decisions without fragmentation
  • 📚 Learning & Research - Maintain context across study sessions with consistent naming
  • ✍️ Writing Projects - Remember plot points without creating duplicate character projects
  • 🔧 Multi-session Debugging - Preserve debugging state with clear project organization

📈 Version History

v1.1.0 (Current)

  • ✨ Project Validation System - Prevents fragmentation with fuzzy name matching
  • ✨ validate_project_name tool - Manual name checking
  • ✨ Force Override capability - Bypass validation when needed
  • ✨ Enhanced Testing - Comprehensive validation test suite
  • 🐛 Bug Fixes - Improved error handling and edge cases

v1.0.0

  • 🚀 Initial release with core continuity functionality

Built with ❤️ for the Claude community

Tired of fragmented projects? Version 1.1 keeps your context organized!

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选