context-killer-mcp
Delegates code generation tasks to Google's Gemini API, writing files directly to disk to save conversation tokens.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@context-killer-mcpwrite a Python script that calculates Fibonacci numbers and save it to fib.py"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Context-Killer MCP
A little MCP server that keeps big code generation out of Claude's context window.
When Claude writes a large file, every line flows back through the conversation and eats tokens. This server hands the job to Google's Gemini instead: Gemini writes the file straight to disk, and Claude only gets back a one-line "done". The code never lands in the chat.
It's a single tool, delegate_to_gemini_and_save. You give it a prompt and a path;
it optionally reads a few files for context, asks Gemini for the code, and saves it.
Only use this for small taks or if you just want to try it out its not fully polished and claude doesnt see the results which is good for saving Tokens but it may lead to Errors since gemini api is not as good in coding and should only be used for small simple Tasks.
The tool
delegate_to_gemini_and_save(task_prompt, target_file_path, context_files=None)
task_prompt– what you want built, in plain wordstarget_file_path– where to save it (missing folders are created)context_files– optional list of files for Gemini to read first
Related MCP server: Claude-to-Gemini MCP Server
Setup
You'll need Python 3.11+ and a free Gemini key from https://aistudio.google.com/app/apikey.
pip install -r requirements.txt
cp .env.example .env # then paste your key into .envUse it in Claude Desktop
Add this to your claude_desktop_config.json:
Windows:
%APPDATA%\Claude\claude_desktop_config.jsonmacOS:
~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"context-killer": {
"command": "python",
"args": ["C:\\path\\to\\context-killer-mcp\\server.py"],
"env": { "GEMINI_API_KEY": "your-key-here" }
}
}
}Restart Claude, then just ask: "Use context-killer to write … and save it to …".
Good to know
It writes (and overwrites) files on disk, so glance at what Claude's about to do before approving.
Your prompt and any context files get sent to Gemini — don't point it at secrets.
.envis git-ignored. Never commit your real key.
License
MIT — see LICENSE.
This server cannot be deployed
Maintenance
Related MCP Connectors
No-data MCP handoff for local Claude Code to Codex harness moves. $49 lifetime.
Give Claude only the Google Drive files you choose. Every action logged.
- SeturosOAuthcom.seturos
Shared work memory for Claude Code, Codex, Cursor and chat, scoped to each repository.
1 Shared memory for coding agents. Stop re-explaining your codebase every session.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceConnects Claude Code with Google's Gemini AI, allowing users to ask Gemini questions, get code reviews, and brainstorm ideas directly within Claude Code.249MIT
- AlicenseBqualityCmaintenanceEnables Claude to use Google Gemini as a secondary AI through MCP for large-scale codebase analysis and complex reasoning tasks. Supports both Gemini Flash and Pro models with specialized functions for general queries and comprehensive code analysis.4MIT
- AlicenseNot gradedqualityDmaintenanceEnables integration with Gemini AI through Claude Code, providing tools to execute queries to Gemini including basic requests and unsafe operations with auto-confirmation.58 npm1ISC
- FlicenseCqualityDmaintenanceEnables Claude Code to call Gemini models through an OpenAI-compatible API, providing tools for deep analysis, brainstorming, code review, and general queries using Gemini's capabilities.4-