ambit-cli（ambit CLI 搜索）

Ambit CLI

What Ambit Is

Ambit deploys apps to the cloud in a way that makes them completely unreachable from the public internet. Only devices enrolled in the user's private Tailscale network can connect to them. From the outside world's perspective, the app does not exist — there is no IP address to find, no port to scan.

This matters because putting a database, dashboard, or internal tool on the normal internet forces you to build login pages, configure firewalls, and harden the app against strangers. Ambit skips all of that. Access control is handled at the network level: if a device is on the Tailscale network, it can reach the app; if it isn't, the connection is refused before the app ever sees it.

Each private network you create is called an ambit. Every app deployed to it gets a human-readable address under that network's name — so http://my-app.lab means the my-app application on the lab ambit. These addresses work automatically for any device enrolled in the user's Tailscale account.

Installation

If ambit is not already installed, run it directly via Nix:

npx @cardelli/ambit

How It Works

graph LR
    A[Your Device<br/>on Tailscale] -->|Tailscale tunnel| B[Ambit Router<br/>Fly.io VM]
    B -->|Flycast / private IPv6| C[Your App<br/>no public IP]

Ambit creates a router on Fly.io that joins the user's Tailscale network and advertises the private IPv6 subnet for that ambit. It also sets up split DNS so that *.<network> queries resolve to the right app. Apps deployed with ambit deploy get a private Flycast address on the network and never receive a public IP.

Prerequisites

• flyctl installed and authenticated (fly auth login)
• Tailscale installed and connected (tailscale up)
• Accept-routes enabled (sudo tailscale set --accept-routes)
• A Tailscale API access token (tskey-api-...) — create one at https://login.tailscale.com/admin/settings/keys

Commands

ambit create <network>

Creates a new private network. This is the first command to run when setting up a new ambit. It deploys a router on Fly.io, connects it to the user's Tailscale network, and configures split DNS so apps on the network are reachable by name.

ambit create lab
ambit create lab --org my-org --region sea
ambit create lab --self-approve

Flags:

• --org <org> — Fly.io organization slug
• --region <region> — Fly.io region (default: iad)
• --api-key <key> — Tailscale API access token (prompted interactively if omitted)
• --tag <tag> — Tailscale ACL tag for the router (default: tag:ambit-<network>)
• --self-approve — Approve subnet routes via Tailscale API instead of requiring autoApprovers in the ACL policy
• -y, --yes — Skip confirmation prompts
• --json — Output as JSON

What it does:

1. Validates Fly.io auth and the Tailscale API key
2. Checks that the tag (default tag:ambit-<network>, or custom via --tag) exists in Tailscale ACL tagOwners
3. Checks autoApprovers config (unless --self-approve)
4. Creates a Fly.io app on the custom network
5. Sets secrets: TAILSCALE_API_TOKEN, NETWORK_NAME, TAILSCALE_TAGS
6. Deploys the router container
7. Waits for the device to join the tailnet
8. Configures split DNS (*.<network> → router)
9. Enables accept-routes locally if possible

Before running, the user must add the router's tag in their Tailscale ACL settings at https://login.tailscale.com/admin/acls/visual/tags. The tag defaults to tag:ambit-<network> but can be overridden with --tag.

And optionally (or use --self-approve to skip this):

"autoApprovers": { "routes": { "fdaa:X:XXXX::/48": ["tag:ambit-<network>"] } }

ambit deploy <app> --network <name>

Deploys an app onto a private network. This is the safe alternative to fly deploy: it always passes --no-public-ips and --flycast, runs pre-flight checks on the fly.toml for dangerous settings, and audits the result to verify no public IPs were allocated.

There are three mutually exclusive deployment modes:

Config mode (default) — uses a local fly.toml:

ambit deploy my-app --network lab
ambit deploy my-app --network lab --config ./custom.toml

Image mode — deploys a Docker image without fly.toml:

ambit deploy my-app --network lab --image registry.fly.io/my-app:latest
ambit deploy my-app --network lab --image registry.fly.io/my-app:latest --main-port 3000

Template mode — fetches a template from a GitHub repository and deploys it:

ambit deploy my-browser --network lab --template ToxicPine/ambit-templates/chromatic
ambit deploy my-browser --network lab --template ToxicPine/ambit-templates/chromatic@v1.0
ambit deploy my-shell --network lab --template ToxicPine/ambit-templates/wetty

Flags:

• --network <name> — Target network (required)
• --org <org> — Fly.io organization
• --region <region> — Primary region
• --config <path> — Explicit path to fly.toml (config mode)
• --image <img> — Docker image to deploy (image mode)
• --main-port <port> — Internal port for HTTP service in image mode (default: 80, "none" to skip)
• --template <ref> — GitHub template reference (template mode)
• -y, --yes — Skip confirmation
• --json — Output as JSON

Template reference format:

owner/repo/path             Fetch from the default branch
owner/repo/path@tag         Fetch a tagged release
owner/repo/path@branch      Fetch a specific branch
owner/repo/path@commit      Fetch a specific commit

The template must contain a fly.toml (and typically a Dockerfile). The template is fetched from GitHub's tarball API, the target subdirectory is extracted, pre-flight scanned, and deployed. The temp directory is cleaned up automatically.

ambit list

Lists all discovered routers across all networks.

ambit list
ambit list --org my-org --json

Shows: network name, app name, region, machine state, private IP, subnet, and Tailscale device status.

ambit status

Shows detailed router status. Run without --network to see all routers, or with --network to focus on one.

ambit status
ambit status --network lab

Detailed view includes: machine state, SOCKS5 proxy address, Tailscale IP, online status, advertised routes, and split DNS config.

ambit destroy --network <name>

Tears down a router and cleans up all associated resources. Apps deployed on the network are NOT deleted — only the router is removed. The user will need to manually remove ACL policy entries for the network tag.

ambit destroy --network lab
ambit destroy --network lab --yes

What it does:

1. Finds the router app for the network
2. Clears split DNS configuration
3. Removes the Tailscale device
4. Destroys the Fly.io app

ambit doctor

Health check for the local environment and router infrastructure. Run this whenever something seems wrong — it checks the most common failure points and gives remediation hints.

ambit doctor
ambit doctor --network lab

Checks:

• Tailscale CLI installed
• Tailscale connected (BackendState = Running)
• Accept-routes enabled
• Router(s) exist and machines are running
• Router(s) visible in tailnet

Templates

Ready-to-deploy templates are available at ToxicPine/ambit-templates:

ambit deploy my-browser --network lab --template ToxicPine/ambit-templates/chromatic
ambit deploy my-shell --network lab --template ToxicPine/ambit-templates/wetty
ambit deploy my-code --network lab --template ToxicPine/ambit-templates/opencode

Common Workflows

First-Time Setup

# 1. Add tag to Tailscale ACL policy in the web UI
# 2. Create the router
ambit create lab --self-approve

# 3. Deploy an app
ambit deploy my-app --network lab

# 4. App is now reachable as http://my-app.lab from any device on the tailnet

# 5. Invite people to your tailnet:
#    https://login.tailscale.com/admin/users
# 6. Control their access:
#    https://login.tailscale.com/admin/acls/visual/general-access-rules

Deploy from a Template

ambit deploy my-browser --network lab --template ToxicPine/ambit-templates/chromatic
# → headless Chrome at my-browser.lab:9222, reachable via CDP

Debugging Connectivity

ambit doctor --network lab    # Check all the common failure points
ambit status --network lab    # Detailed router state

Tearing Down

ambit destroy --network lab
# Then remove from Tailscale ACL:
#   tagOwners: tag:ambit-lab
#   autoApprovers: routes for tag:ambit-lab

Troubleshooting