Skip to main content

Smart MCP profile manager for Claude Code — save 30-40% of your context window

Project description

MCPSwitch

Smart MCP profile manager for Claude Code.
Stop wasting 30-40% of your context window on MCP tools you never use.


The Problem

Every MCP server you install adds its tool schemas to Claude's context window at session start.
84 tools across 6 servers = 15,540 tokens consumed before you type a single message.
That's up to 40% of your context window gone — every session, every time.

MCPSwitch lets you define lean profiles for different workflows and switch between them instantly.


Install

pip install mcpswitch

Or run from source:

git clone https://github.com/sathibabu/mcpswitch
cd mcpswitch
pip install -e .

Quick Start

# See what your current MCP config is costing you
mcpswitch analyze

# Save your current config as a profile
mcpswitch import --name full

# Create a lean profile
mcpswitch create python-backend

# Add only the servers you need for Python work
mcpswitch add python-backend github
mcpswitch add python-backend context7

# Switch to it before starting a Claude Code session
mcpswitch use python-backend

# See all profiles and their token costs
mcpswitch list

Commands

Command What it does
mcpswitch status Show active profile + token cost of loaded servers
mcpswitch analyze Break down token cost by server, show savings potential
mcpswitch list List all profiles with token estimates
mcpswitch use <profile> Switch to a profile (rewrites Claude config)
mcpswitch import --name <n> Save current config as a named profile
mcpswitch create <profile> Create a new empty profile
mcpswitch add <profile> <server> Add a server to a profile
mcpswitch remove <profile> <server> Remove a server from a profile
mcpswitch save <profile> Save current active config as profile
mcpswitch delete <profile> Delete a profile

Real-World Savings

Scenario All MCPs Lean Profile Saved
Python backend work 15,540 tokens 3,200 tokens 12,340 tokens (79%)
Frontend work 15,540 tokens 4,100 tokens 11,440 tokens (74%)
Writing/docs 15,540 tokens 900 tokens 14,640 tokens (94%)

Fewer wasted tokens = more context for your actual code = better Claude responses.


How It Works

MCPSwitch reads and writes the Claude MCP config files:

  • Claude Code CLI: ~/.claude/claude_desktop_config.json
  • Claude Desktop: ~/AppData/Roaming/Claude/claude_desktop_config.json (Windows)

Profiles are stored in ~/.mcpswitch/profiles.json. Every config change is backed up automatically before overwriting.


Pricing

  • Free: unlimited profiles, all commands, open source
  • Pro ($19/month): coming soon — auto-detect project type and switch profiles automatically, usage analytics, team profile sharing

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mcpswitch_cli-0.1.0.tar.gz (39.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

mcpswitch_cli-0.1.0-py3-none-any.whl (44.6 kB view details)

Uploaded Python 3

File details

Details for the file mcpswitch_cli-0.1.0.tar.gz.

File metadata

  • Download URL: mcpswitch_cli-0.1.0.tar.gz
  • Upload date:
  • Size: 39.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.0

File hashes

Hashes for mcpswitch_cli-0.1.0.tar.gz
Algorithm Hash digest
SHA256 6447ef6792de54d5f26a9389bd4b8b96cb6db20a3e4deeafecd518121e15bbe2
MD5 fe8890e7ecfa75c6607400415b1272c0
BLAKE2b-256 106870722d8cb1d830c1ac027533b6a5db3edcd4c4daafbb9e6b32662de2a220

See more details on using hashes here.

File details

Details for the file mcpswitch_cli-0.1.0-py3-none-any.whl.

File metadata

  • Download URL: mcpswitch_cli-0.1.0-py3-none-any.whl
  • Upload date:
  • Size: 44.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.0

File hashes

Hashes for mcpswitch_cli-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 16c68d06a29c8f86814d7987a86ec6cf6a8ec540c508c0befdf8d11060194bff
MD5 463def078d5eba555cba545b932df7d8
BLAKE2b-256 a74a442b551754c8333a219a51bc0b4fe33be5a247db061c415ae080df0114df

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page