dotfiles-manager MCP Server
Enables AI agents to manage dotfiles using git bare repos, supporting tracking, syncing, multiple stores, and remote setup through natural language.
README
dotfiles-manager
A thin, open-source wrapper around git's bare-repo dotfiles pattern. It reduces the cognitive load of git plumbing without hiding it. Your files stay where they are, tracked in standard git repos — no new abstractions, no proprietary formats, no lock-in.
If you stop using this tool tomorrow, your dotfiles and their full history
are still right there in git. You can drop to raw git commands at any time
via dot git, or walk away entirely with nothing to unwind.
See Alternatives for a detailed comparison with the manual bare-repo approach and chezmoi.
Why
- One command to back up:
dot synccommits and pushes everything - One command to set up a new machine:
dot remote setup+dot sync - Full version history: Roll back any file with
dot git log - No new mental model: It's git. If you know git, you know everything
- Works without GitHub: Local-only mode until you're ready to push
- Multiple stores: Manage independent repos from one CLI with
--store - Safety First: Dangerous commands require explicit store confirmation
- AI-agent friendly: MCP server lets Claude or other assistants manage your dotfiles
Architecture
All business logic lives in the SDK (dotgit/sdk/). The CLI and MCP server
are thin interface layers — they parse arguments, call SDK functions, and
format output. No business logic in CLI or MCP.
dotgit/
sdk/ ← all logic: tracking, syncing, remote setup, hooks, stores
cli/ ← thin CLI wrapper (calls SDK, formats for terminal)
mcp/ ← thin MCP wrapper (calls SDK, exposes as tool schema)
See CONTRIBUTING.md for architecture and testing details.
Install
pipx install git+https://github.com/krisrowe/dotfiles-manager.git
This gives you two commands: dot (CLI) and dot-mcp (MCP server).
User Journeys
First machine: start tracking dotfiles
# Set the active store for the current machine
dot default personal
# Track files (requires explicit --store for safety)
dot --store=personal track ~/.bashrc
dot --store=personal track ~/.config/myapp/
# Setup remote and sync (can use active store implicitly)
dot remote setup --repo-name my-dotfiles
dot sync
Each dot track commits immediately to a local bare repo (e.g. ~/.dotfiles-personal).
dot remote setup creates a private GitHub repo and sets a
dotfiles-personal topic for discovery. dot sync pushes.
New machine: discover and set up existing stores
# Discover available dotfiles on your GitHub account
dot remote available
# Setup a discovered store
dot --store=personal remote setup --repo-name my-dotfiles
dot default personal
dot sync
remote setup attaches to the existing repo. sync pulls everything down and checks out
tracked files to their original paths.
Day-to-day: back up changes
dot sync
Commits any changes, pulls from GitHub, pushes to GitHub. Uses the active store configured for the current machine.
Multiple stores: separate repos for different purposes
# Create a second store
dot stores create work
# Track files scoped to the store (REQUIRED flag)
dot --store=work track ~/.config/work-app/settings.conf
# Sync specifically for the work store
dot --store=work sync
# Or switch your active store and sync normally
dot default work
dot sync
Commands
Day-to-day
| Command | Description |
|---|---|
dot track <path> |
Start tracking (Requires --store) |
dot untrack <path> |
Stop tracking (Requires --store) |
dot list |
List all tracked files in active store |
dot status |
Show modified files in active store |
dot sync |
Commit + pull + push active store |
dot default [NAME] |
View or set the machine's active store |
Exclude patterns
Control what gets ignored inside tracked directories.
| Command | Description |
|---|---|
dot exclude add <pattern> |
Add a gitignore-format pattern |
dot exclude remove <pattern> |
Remove a pattern |
dot exclude list |
Show current patterns |
Remote management
| Command | Description |
|---|---|
dot remote setup [--repo-name NAME] |
Create/verify private GitHub repo |
dot remote show |
Show remote URL of active store |
dot remote available |
Discover available stores on GitHub |
Git hooks
Your global git hooks run on dotfile commits by default. If they interfere:
| Command | Description |
|---|---|
dot hooks disable |
Disable hooks on current store (persistent) |
dot hooks reset |
Restore global hooks |
dot hooks show |
Show current hooks state |
Stores
| Command | Description |
|---|---|
dot stores create <name> |
Create a new store at ~/.dotfiles-<name> |
dot stores list |
Show all stores (* = active) |
Pass-through git
| Command | Description |
|---|---|
dot git <args> |
Run git commands (Requires --store) |
dot --store=work git log --oneline -10
dot --store=personal git diff HEAD~1
Multiple Stores
Manage independent bare repos from one CLI with --store. Each store gets
its own tracking, hooks configuration, git remote, and backup strategy.
--store is a top-level option. When omitted, the active store configured via dot default is used for safe commands (sync, status), while risky commands (track, git) will require an explicit flag to prevent accidental cross-contamination.
MCP Server
The MCP server exposes the same SDK functions as the CLI. All MCP tools accept an optional store parameter.
Register with Claude Code
dot mcp install claude
Lay of the Land (Discovery Journey)
Managing a home directory as a work tree can be overwhelming. dotgit provides a structured journey to help you get a lay of the land without the noise.
1. What am I already backing up?
Start with dot list to see exactly which files and directories are currently tracked in your active store.
dot list
2. What needs to be synced?
Use dot status to see local modifications to your tracked files. This is your primary "to-do" list before running dot sync.
dot status
3. What dotfiles am I missing?
dot status also automatically surfaces hidden files and folders (.*) in your home directory that aren't tracked in any of your registered stores. This is the fastest way to discover new configuration files that should be backed up.
4. How do I verify my ignore rules?
If you want to see which hidden files or folders you have intentionally excluded via your ~/.config/git/ignore rules, use the --ignored flag:
dot status --ignored
This is useful for auditing your "black boxes"—ensuring that directories you once chose to forget (like .npm/ or .cache/) are still correctly excluded.
Pro-Tip: Visibility & "Explicit Siblings" If a directory appears in the Ignored list but also contains files that are Tracked (
dot list), Modified (dot status), or Discovered (Untracked), it is a "Partially Open Box." This happens when you ignore a parent directory but explicitly track or add an exception (!) for a specific child.The Visibility Risk: Ignoring a parent directory creates a blackout. If new, valuable files are added to that folder later, Git will silently swallow them, and you will never be prompted to make a conscious choice about tracking them.
The Strategy: For low-noise directories that contain valuable settings (like
.gemini/), we recommend ignoring explicit siblings (the specific files and/or sub-folders you don't want) instead of the parent. This keeps the parent "open" so that any new, important files will surface indot statusfor your review.
How It Works
...
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.