Tailscale skill

A skill that gives your coding agent deep knowledge of Tailscale and the wider Tailscale product family. It helps you set up an exit node on your home server, wire up Aperture for LLM API governance, configure the Kubernetes operator, develop applications with tsnet, and everything in between.

This skill works with Claude Code, Cursor, OpenCode, OpenAI Codex, Pi, and any other agent that follows the Agent Skills standard.

Note

This skill is a public Alpha release and is not yet supported. Please don't file Tailscale support tickets for issues you discover with it, and treat its output as a starting point rather than a final answer. Configuration syntax in particular evolves; the skill links out to canonical Tailscale documentation for the parts that change most often.

Refer to Tailscale release stages for more information about alpha, beta, and GA releases.

Installing

Install using the skills CLI with npx:

npx skills add https://github.com/tailscale/tailscale-skill

Using the Tailscale skill

You activate the skill two ways:

• Explicit: type /tailscale followed by your question. This always invokes the skill when installed.
• Implicit: ask your LLM a Tailscale question conversationally, like "how do I set up a subnet router for my home network?", and the LLM consults the skill if the question matches the skill's description.

The implicit path depends on your LLM's judgment and isn't 100% reliable. When in doubt, use the /tailscale slash command.

What's in the skill

SKILL.md is a topic index that routes the LLM to focused reference files in the references/ directory. Topics include:

• Core mesh VPN: exit nodes, subnet routers, MagicDNS, grants (and legacy ACLs), tagging
• Connectivity diagnostics: DERP, NAT traversal, peer relay, tailscale netcheck / ping / status
• Sharing & publishing: Taildrop, Taildrive, Tailscale Serve (private), Tailscale Funnel (public)
• Containers & Kubernetes: the operator, Docker sidecar pattern, ProxyGroup, Connector CRD, API server proxy
• Enterprise rollout: SCIM provisioning, MDM (Jamf/Intune), device posture, device approval, auth keys, SSO
• Session recording: tsrecorder setup, S3 destinations, fail-closed enforcement, SSH and Kubernetes recording
• Aperture: AI gateway for centralizing LLM API keys, quotas, per-user usage dashboards, coding-agent integration
• CLI: every tailscale subcommand with flags and examples
• REST API: authentication, device management, policy file, webhooks
• Building Go applications on Tailscale: using tsnet to build applications that embed the Tailscale client.

Each reference file is self-contained. The LLM should only load the relevant references to answer your question.

Issues and feedback

Open an issue at https://github.com/tailscale/tailscale-skill/issues. This is the place for skill-specific problems (a reference is wrong, your agent gives bad advice for some feature, the install instructions don't work).

For Tailscale product questions unrelated to the skill, use the regular Tailscale support channels.

Legal

WireGuard is a registered trademark of Jason A. Donenfeld.

LICENSE

BSD 3-Clause "New" or "Revised" License. Refer to LICENSE for details.