Sync AI assistant bundles — skills, rules, agents, slash commands — across projects, machines, and tools from a single registry.
- You tweak a skill in
~/.claude/on WSL — your Windows host still has the old copy. - You build a slash command for a Claude project, then open the next one in Cursor and want the same thing, in Cursor's format.
- Copy-pasting bundles between machines and tools gets old fast.
Point every project at one registry (a local folder or git repo). Bundles are stored once in canonical Claude format; pull transforms them on the way out into whatever the project's tool expects (.claude/ or .cursor/). Edit locally, push back, and every other project picks it up on the next sync.
skillsync init # set up this project (and optionally wire a git hook)
skillsync pull # fetch from registry → ./.claude/ or ./.cursor/
skillsync push # send local edits back to the registry
skillsync sync # both directions, with conflict detectionOpt into the git pre-commit hook during init and every commit keeps your registry in lockstep.
Linux / macOS — installs to ~/.local/bin:
curl -fsSL https://raw.githubusercontent.com/rolfwessels/skillsync/main/install.sh | shWindows (PowerShell) — installs to %LOCALAPPDATA%\Programs\skillsync\ and adds it to your user PATH:
irm https://raw.githubusercontent.com/rolfwessels/skillsync/main/install.ps1 | iexWant a different location? Set INSTALL_DIR first:
INSTALL_DIR=/usr/local/bin curl -fsSL https://raw.githubusercontent.com/rolfwessels/skillsync/main/install.sh | sh$env:INSTALL_DIR = 'C:\tools\skillsync'; irm https://raw.githubusercontent.com/rolfwessels/skillsync/main/install.ps1 | iexTo upgrade, just re-run the same command. Both scripts pull the binary from the latest GitHub release, which is published on every push to main under a new version tag.
Prefer to download by hand? Grab the right archive for your platform from the releases page, extract, and put the binary somewhere on your PATH.
- Cobra for the CLI
- Bubble Tea + Huh for the TUI
- BurntSushi/toml for config parsing
- Docker for the dev environment
- MakeFile because it just works!
This project ships with a development container that has all the tooling required to build, test, and publish.
# bring up dev environment
make build up
# test the project
make test
# run the CLI
make start
# build release binaries for all platforms
make publishTo build and push a Docker image:
make docker-build docker-login docker-push
# or just
make docker-publish- Docker — for the dev container
- Git — for version control
make— available via WSL on Windows, or natively on macOS/Linux
| Command | Description |
|---|---|
make up |
Bring up the container & attach to the dev shell |
make down |
Stop the container |
make build |
Rebuild the container |
| Command | Description |
|---|---|
make version |
Show the current version |
make start |
Run skillsync |
make test |
Run tests |
make publish |
Build release archives for 5 platforms |
make docker-login |
Login to Docker registry |
make docker-build |
Build the production Docker image |
make docker-push |
Push the Docker image |
make docker-pull-short-tag |
Pull image by git short hash |
make docker-tag-env |
Tag image for an environment |
make docker-publish |
Full build + push workflow |
make deploy |
Deploy skillsync |
make update-packages |
Update Go dependencies to latest |
This project follows Semantic Versioning:
- MAJOR: Incompatible API changes
- MINOR: Backward-compatible new functionality
- PATCH: Backward-compatible bug fixes
MAJOR and MINOR are set manually via versionPrefix in the Makefile. PATCH is automatically derived from commit count.
make versionFeature branches are created off main with the prefix feature/ or bug/.
PR builds attach archives as workflow artifacts. Merging to main publishes them as a new versioned GitHub release.
Can I use this on Windows/macOS/Linux?
Yes — binaries are published for all three platforms.
How do I update to the latest version?
Re-run the install command (install.sh or install.ps1). It overwrites the binary in place from the latest release.