Your dev servers, on your tailnet.
tailserve watches the development servers running on your machine and makes the ones you allow reachable from every device in your Tailscale network. Automatically, with a dashboard that links to all of them.
🤔 Why?
If you use Tailscale, you’ve probably used tailscale serve to access a local dev server from your phone. I did too.
The routine was always the same: npm run dev, then tailscale serve --bg 5173, then grab my phone. Then I’d start another project on 5174, maybe a Storybook on 6006, and repeat the whole thing.
Eventually, I’d stop everything but forget to clean up the Tailscale mappings. The next day, I’d find a bunch of stale entries pointing to ports where nothing was running anymore.
Don’t get me wrong, I love Tailscale Serve. I often work remotely, sometimes from public networks, so being able to access my local dev servers from my other devices without exposing them to the internet is something I really appreciate. I just got tired of managing the mappings myself.
So I wrote a tool that does it for me.
Start a dev server, and it shows up on your tailnet. Stop it, and the mapping disappears.
That’s it. That’s tailserve.
🚀 Quickstart
You need macOS and Tailscale logged in to a tailnet.
# install
brew install marcomontalbano/tap/tailserve
# start a dev server as usual, then pick what to serve
tailserve allow
# serve it
tailserve start
tailserve start prints the address of the dashboard, something like https://my-laptop.your-tailnet.ts.net/. Open it from any device on your tailnet 😎
Note
Not using Homebrew? Download the archive for your Mac from the latest release, or build it with go install github.com/marcomontalbano/tailserve@latest.
✨ In Short
- Nothing is served until you allow it. You decide with a few Rules: a port, a range of ports, a process name like
node, or a port and a process together. - It follows your dev servers. Every 2 seconds tailserve looks at what’s listening. When an allowed server starts, it’s served; when it stops (or crashes), it’s gone.
- Sensitive ports are locked. SSH, databases, debuggers like the Node inspector and screen sharing are never served by a broad Rule, only by one that names that exact port.
- Nothing is left behind. Everything tailserve serves lives in its own Tailscale session, so stopping tailserve (or a crash) takes it all offline.
- It never touches what you served yourself. Mappings you created with
tailscale serveare shown, never changed. - A dashboard at your machine’s tailnet address, and at http://localhost:7171 on the machine itself, with a link to every server, grouped by repository.
📖 Usage
tailserve start # serve the allowed Services until stopped
tailserve stop # stop the running tailserve; everything it served goes offline
tailserve status # what's allowed or served, and why
tailserve allow # pick running Services to serve, or add a Rule directly
tailserve deny # pick Rules to remove
tailserve config # open the config file in $EDITOR
tailserve version # the version of tailserve; add --daemon for the running one too
Allow and deny
Run tailserve allow without flags and it asks you which of the running servers to serve, and how the Rule should match each one:
- port
5173, only whennodelistens on it - port
5173, whatever process listens on it - any
nodeprocess, on any port
Or add a Rule directly:
tailserve allow --port 5173-5199 --note vite # a range of ports
tailserve allow --process node # every node process
tailserve allow --port 8000 --process python3 # both must match
tailserve deny --process node
A running tailserve picks up the change within 2 seconds.
Status
tailserve status tells you, for each server, whether it’s served and why:
PORT TOOL REPOSITORY PID STATE DETAIL
4321 astro dev (node) marcomontalbano.github.io 52647 served https://my-laptop.your-tailnet.ts.net:4321
5180 storybook dev (node) tailserve 46138 not allowed no Rule matches
9229 node tailserve 46139 locked port 9229 is locked: only a Rule naming exactly this port serves it
Add --all to also list the servers no Rule matches.
⚙️ Config
The config lives in ~/.config/tailserve/config.toml. tailserve allow and tailserve deny write it for you; tailserve config opens it if you prefer editing by hand.
[[allow]]
port = "5173-5199" # a port or a range of ports
process = "node" # a process name; with a port too, both must match
note = "vite" # a reminder of what the Rule is for
[locked]
ports = [4000] # more locked ports: only a Rule naming one exactly serves it
[dashboard]
local_port = 7171 # the dashboard at http://localhost:7171; 0 turns it off
Locked ports
These ports are never served by a range or a process name, only by a Rule that names that exact port:
22 SSH · 139 445 SMB · 548 AFP · 2345 Delve · 3306 MySQL · 3389 Remote Desktop · 5432 PostgreSQL · 5900 VNC · 6379 Redis · 9229 Node inspector · 27017 MongoDB
Note
Ports 443 and 80 on the tailnet are taken by the dashboard and its redirect.
🛠️ Development stuff
Requirements
Setup
git clone https://github.com/marcomontalbano/tailserve.git
cd tailserve
go test ./... # run the tests
go run . status # run the CLI from the sources
go build -o bin/tailserve . # development build
go build -trimpath -ldflags="-s -w" -o bin/tailserve . # smaller build, for releases
gofmt -l . # list the files that aren't formatted (no output means all good)
go fmt ./... # format them
go vet ./... # catch suspicious code the compiler accepts
# check the dependencies for known vulnerabilities
go run golang.org/x/vuln/cmd/govulncheck@latest ./...
To try a different config without touching yours, point XDG_CONFIG_HOME at another folder:
XDG_CONFIG_HOME=/tmp/tailserve-dev go run . status # reads /tmp/tailserve-dev/tailserve/config.toml