Sandbox
@mcs-cli/mcs

CLI for packaged Claude Code environments

mcs is a configuration engine for Claude Code that keeps agent setup reproducible. You add tech packs, then `mcs sync` installs the right artifacts into a project or global environment and tracks changes over time. It also provides `mcs doctor` for verification and `mcs pack` commands for building and validating packs.

92 starsโ€ข8 forksโ€ขSwiftโ€ขUpdated 7d ago
Who it's for

Builders who want to share, copy, and keep Claude Code setups in sync across machines and projects.

What it delivers

You can rebuild a Claude Code environment in minutes instead of redoing MCP servers, hooks, skills, and settings by hand.

What it does

Tech pack sync

Installs pack contents into the right project or global locations with `mcs sync`.

Pack management

Adds, exports, validates, and updates tech packs with `mcs pack` commands.

Drift detection and repair

`mcs doctor` checks the setup and `mcs doctor --fix` repairs missing or changed pieces.

Reproducible configuration

Lockfiles and pack manifests pin versions so the same environment can be recreated later.

Safety features

Uses dry runs, backups, and trust checks before changing user files.

How to get it

  1. 1Run
    brew install mcs-cli/tap/mcs
  2. 2Browse available packs at techpacks.mcs-cli.dev, then add them
    mcs pack add owner/pack-name
  3. 3Run
    cd ~/Developer/my-project
    mcs sync
  4. 4Run
    mcs doctor

README

mcsmcs

Your Claude Code environment โ€” packaged, portable, and reproducible.

Swift 6.0+ macOS License: MIT Homebrew Tech Packs Ask DeepWiki

๐Ÿš€ Quick Start

1. Install

brew install mcs-cli/tap/mcs

2. Add tech packs

Browse available packs at techpacks.mcs-cli.dev, then add them:

mcs pack add owner/pack-name

3. Sync a project

cd ~/Developer/my-project
mcs sync

4. Verify everything

mcs doctor

That's it. Your MCP servers, plugins, hooks, skills, commands, agents, settings, and templates are all in place.


The Problem

You've spent hours getting Claude Code just right โ€” MCP servers, plugins, hooks, skills, custom commands, fine-tuned settings. Then:

  • ๐Ÿ–ฅ๏ธ New machine? Start over from scratch.
  • ๐Ÿ‘ฅ Onboarding a teammate? "Just follow this 47-step wiki page."
  • ๐Ÿ“‚ Different projects? Copy-paste configs, hope nothing drifts.
  • ๐Ÿ”„ Something broke? Good luck figuring out what changed.

The Solution

mcs is a configuration engine for Claude Code โ€” like Ansible for your AI development environment. Package everything into shareable tech packs (Git repos with a techpack.yaml manifest), then sync them across any project, any machine, in seconds.

Without mcsWith mcs
Install MCP servers one by onemcs pack add + mcs sync
Hand-edit settings.json per projectManaged settings composition
Copy hooks between projects manuallyAuto-installed per-project from packs
Configuration drifts silentlymcs doctor --fix detects and repairs
Rebuild from memory on new machinesFully reproducible in minutes
No way to share your setupPush a pack, anyone can mcs pack add it

๐Ÿ” Explore Tech Packs

Packs are modular โ€” mix and match what you need instead of one monolith. Browse the full catalog, search by category, and find install commands:

techpacks.mcs-cli.dev

๐Ÿ’ก Can't find what you need? Build your own โ€” see Creating Tech Packs.


โš™๏ธ How It Works

 Tech Packs          mcs sync          Your Project
 (Git repos)  -----> (engine)  -----> (configured)
                        |
                   .---------.
                   |         |
                   v         v
              Per-Project  Global
              artifacts    artifacts
  1. Select which packs to apply (interactive multi-select or --all)
  2. Resolve prompts (auto-detect project files, ask for config values)
  3. Install artifacts to the right locations (skills, hooks, commands, agents, settings, MCP servers)
  4. Track everything for convergence โ€” re-running mcs sync adds what's missing, removes what's deselected, and updates what changed

Use mcs sync --global for global-scope components (Homebrew packages, plugins, global MCP servers). See Architecture for artifact locations and the full sync flow.


๐Ÿ“ฆ What's in a Tech Pack?

A tech pack is a Git repo with a techpack.yaml manifest. It can bundle MCP servers, plugins, hooks, skills, commands, agents, settings, templates, and doctor checks โ€” anything mcs can install, verify, and uninstall.

Use mcs pack validate to check your pack for structural errors and best practices before sharing.

๐Ÿ“– Full guide: Creating Tech Packs ยท Schema reference: techpack-schema.md


๐ŸŽฏ Use Cases

  • ๐Ÿง‘โ€๐Ÿ’ป Solo Developer โ€” New Mac? One mcs pack add + mcs sync and your entire Claude Code environment is back. No wiki, no notes, no memory required.
  • ๐Ÿ‘ฅ Teams โ€” Create a team pack with your org's MCP servers, approved plugins, and coding standards. Every developer gets the same setup with mcs sync --all.
  • ๐ŸŒ Open Source โ€” Use mcs export to create a tech pack from your repo's config. Contributors run mcs sync and get the right MCP servers, skills, and conventions automatically.
  • ๐Ÿงช Experimentation โ€” Try a different set of MCP servers, swap packs, roll back. mcs converges cleanly โ€” deselected packs are fully removed, no leftovers.

๐Ÿ›ก๏ธ Safety & Trust

mcs is designed to be non-destructive and transparent. Timestamped backups before modifying user content, --dry-run to preview changes, section markers to preserve your edits in CLAUDE.local.md, and SHA-256 trust verification for pack scripts. Lockfiles (mcs.lock.yaml) pin pack versions for reproducible environments.

๐Ÿ“– Full details: Architecture > Safety & Trust


๐Ÿ” Verifying Your Setup with Poirot

After mcs sync, want to confirm everything landed correctly? Poirot is a native macOS companion that gives you a visual overview of your Claude Code configuration โ€” MCP servers, settings, sessions, and more โ€” all in one place.

The perfect complement to mcs: configure your environment with mcs, then use Poirot to see exactly what's installed and running.


๐Ÿ“š Documentation

DocumentDescription
๐Ÿ“– CLI ReferenceComplete command reference (sync, pack, doctor, export, cleanup, check-updates, config)
๐Ÿ“– Creating Tech PacksStep-by-step guide to building your first pack
๐Ÿ“‹ Tech Pack SchemaComplete techpack.yaml field reference
๐Ÿ—๏ธ ArchitectureInternal design, sync flow, safety guarantees, and extension points
๐Ÿ”ง TroubleshootingCommon issues and fixes

๐Ÿ› ๏ธ Development

swift build                                            # Build
swift test                                             # Run tests
swift build -c release --arch arm64 --arch x86_64      # Universal binary

See Architecture for project structure and design decisions.

๐Ÿค Contributing

Tech pack ideas and engine improvements are welcome!

  1. Fork the repo
  2. Create a feature branch
  3. Run swift test
  4. Open a PR

For building new packs, start with Creating Tech Packs.


๐Ÿ’› Support

If mcs saves you time, consider sponsoring the project.

MIT License ยท Made with โค๏ธ by Bruno Guidolim ยท mcs-cli

Files in the repo

Repository payloadโ€ข12 top-level entries
  • .github
  • docs
  • skills
  • Sources
  • Tests
  • .gitignore
  • .swiftformat
  • .swiftlint.yml
  • CLAUDE.md
  • LICENSE
  • Package.swift
  • README.md

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More tools

JuliusBrussee/
caveman

๐Ÿชจ why use many token when few token do trick โ€” Claude Code skill that cuts 65% of tokens by talking like caveman

105k
1 add
MemPalace/
mempalace

The best-benchmarked open-source AI memory system. And it's free.

59k
stablyai/
orca

Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

66k

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

132k

Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors

64k
headroomlabs-ai/
headroom

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

71k