Skip to content
Back to search

Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP

🧠 MCP server implementing RAT (Retrieval Augmented Thinking) - combines DeepSeek's reasoning with GPT-4/Claude/Mistral responses, maintaining conversation context between interactions.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/newideas99--deepseek-thinking-claude-3-5-sonnet-cline-mcp/badge.svg)](/listings/newideas99--deepseek-thinking-claude-3-5-sonnet-cline-mcp)

HTML

<a href="/listings/newideas99--deepseek-thinking-claude-3-5-sonnet-cline-mcp"><img src="/api/listings/newideas99--deepseek-thinking-claude-3-5-sonnet-cline-mcp/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Installing via Smithery

To install DeepSeek Thinking with Claude 3.5 Sonnet for Claude Desktop automatically via [Smithery](https://smithery.ai/server/@newideas99/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP):

npx -y @smithery/cli install @newideas99/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP --client claude

Manual Installation

1. Clone the repository:

git clone https://github.com/yourusername/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP.git
cd Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP

2. Install dependencies:

npm install

3. Create a `.env` file with your OpenRouter API key:

# Required: OpenRouter API key for both DeepSeek and Claude models
OPENROUTER_API_KEY=your_openrouter_api_key_here

# Optional: Model configuration (defaults shown below)
DEEPSEEK_MODEL=deepseek/deepseek-r1  # DeepSeek model for reasoning
CLAUDE_MODEL=anthropic/claude-3.5-sonnet:beta  # Claude model for responses

4. Build the server:

npm run build

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 184 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP
  2. [![smithery badge](https://smithery.ai/badge/@newideas99/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP)](https://smithery.ai/server/@newideas99/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP)
  3. A Model Context Protocol (MCP) server that combines DeepSeek R1's reasoning capabilities with Claude 3.5 Sonnet's response generation through OpenRouter. This implementation uses a two-stage process where DeepSeek provides structured reasoning which is then incorporated into Claude's response generation.
  4. ## Features
  5. - **Two-Stage Processing**:
  6. - Uses DeepSeek R1 for initial reasoning (50k character context)
  7. - Uses Claude 3.5 Sonnet for final response (600k character context)
  8. - Both models accessed through OpenRouter's unified API
  9. - Injects DeepSeek's reasoning tokens into Claude's context
  10. - **Smart Conversation Management**:
  11. - Detects active conversations using file modification times
  12. - Handles multiple concurrent conversations
  13. - Filters out ended conversations automatically
  14. - Supports context clearing when needed
  15. - **Optimized Parameters**:
  16. - Model-specific context limits:
  17. * DeepSeek: 50,000 characters for focused reasoning
  18. * Claude: 600,000 characters for comprehensive responses
  19. - Recommended settings:
  20. * temperature: 0.7 for balanced creativity
  21. * top_p: 1.0 for full probability distribution
  22. * repetition_penalty: 1.0 to prevent repetition
  23. ## Installation
  24. ### Installing via Smithery
  25. To install DeepSeek Thinking with Claude 3.5 Sonnet for Claude Desktop automatically via [Smithery](https://smithery.ai/server/@newideas99/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP):
  26. ```bash
  27. npx -y @smithery/cli install @newideas99/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP --client claude
  28. ```
  29. ### Manual Installation
  30. 1. Clone the repository:
  31. ```bash
  32. git clone https://github.com/yourusername/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP.git
  33. cd Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP
  34. ```
  35. 2. Install dependencies:
  36. ```bash
  37. npm install
  38. ```
  39. 3. Create a `.env` file with your OpenRouter API key:
  40. ```env
  41. # Required: OpenRouter API key for both DeepSeek and Claude models
  42. OPENROUTER_API_KEY=your_openrouter_api_key_here
  43. # Optional: Model configuration (defaults shown below)
  44. DEEPSEEK_MODEL=deepseek/deepseek-r1 # DeepSeek model for reasoning
  45. CLAUDE_MODEL=anthropic/claude-3.5-sonnet:beta # Claude model for responses
  46. ```
  47. 4. Build the server:
  48. ```bash
  49. npm run build
  50. ```
  51. ## Usage with Cline
  52. Add to your Cline MCP settings (usually in `~/.vscode/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json`):
  53. ```json
  54. {
  55. "mcpServers": {
  56. "deepseek-claude": {
  57. "command": "/path/to/node",
  58. "args": ["/path/to/Deepseek-Thinking-Claude-3.5-Sonnet-CLINE-MCP/build/index.js"],
  59. "env": {
  60. "OPENROUTER_API_KEY": "your_key_here"
  61. },
  62. "disabled": false,
  63. "autoApprove": []
  64. }
  65. }
  66. }
  67. ```
  68. ## Tool Usage
  69. The server provides two tools for generating and monitoring responses:
  70. ### generate_response
  71. Main tool for generating responses with the following parameters:
  72. ```typescript
  73. {
  74. "prompt": string, // Required: The question or prompt
  75. "showReasoning"?: boolean, // Optional: Show DeepSeek's reasoning process
  76. "clearContext"?: boolean, // Optional: Clear conversation history
  77. "includeHistory"?: boolean // Optional: Include Cline conversation history
  78. }
  79. ```
  80. ### check_response_status
  81. Tool for checking the status of a response generation task:
  82. ```typescript
  83. {
  84. "taskId": string // Required: The task ID from generate_response
  85. }
  86. ```
  87. ### Response Polling
  88. The server uses a polling mechanism to handle long-running requests:
  89. 1. Initial Request:
  90. - `generate_response` returns immediately with a task ID
  91. - Response format: `{"taskId": "uuid-here"}`
  92. 2. Status Checking:
  93. - Use `check_response_status` to poll the task status
  94. - **Note:** Responses can take up to 60 seconds to complete
  95. - Status progresses through: pending → reasoning → responding → complete
  96. Example usage in Cline:
  97. ```typescript
  98. // Initial request
  99. const result = await use_mcp_tool({
  100. server_name: "deepseek-claude",
  101. tool_name: "generate_response",
  102. arguments: {
  103. prompt: "What is quantum computing?",
  104. showReasoning: true
  105. }
  106. });
  107. // Get taskId from result
  108. const taskId = JSON.parse(result.content[0].text).taskId;
  109. // Poll for status (may need multiple checks over ~60 seconds)
  110. const status = await use_mcp_tool({
  111. server_name: "deepseek-claude",
  112. tool_name: "check_response_status",
  113. arguments: { taskId }
  114. });
  115. // Example status response when complete:
  116. {
  117. "status": "complete",
  118. "reasoning": "...", // If showReasoning was true
  119. "response": "..." // The final response
  120. }
  121. ```
  122. ## Development
  123. For development with auto-rebuild:
  124. ```bash
  125. npm run watch
  126. ```
  127. ## How It Works
  128. 1. **Reasoning Stage (DeepSeek R1)**:
  129. - Uses OpenRouter's reasoning tokens feature
  130. - Prompt is modified to output 'done' while capturing reasoning
  131. - Reasoning is extracted from response metadata
  132. 2. **Response Stage (Claude 3.5 Sonnet)**:
  133. - Receives the original prompt and DeepSeek's reasoning
  134. - Generates final response incorporating the reasoning
  135. - Maintains conversation context and history
  136. ## License
  137. MIT License - See LICENSE file for details.
  138. ## Credits
  139. Based on the RAT (Retrieval Augmented Thinking) concept by [Skirano](https://x.com/skirano/status/1881922469411643413), which enhances AI responses through structured reasoning and knowledge retrieval.
  140. This implementation specifically combines DeepSeek R1's reasoning capabilities with Claude 3.5 Sonnet's response generation through OpenRouter's unified API.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.