Publish from your AI
Connect Publidock once and publish from the conversation you’re already having. Your assistant hands back the address and tells you who can open it. The CLI, GitHub Action and HTTP API do the same from scripts.
What it does
- One connector, no key to paste: Add Publidock to Claude, ChatGPT, Cursor, VS Code, Codex, Gemini CLI or any MCP client, then sign in when asked.
- It can do the whole job: Publish a site, update it, set who can open it, list your sites, and send work to a canvas.
- You stay in charge: You choose its permissions, it tells you who can open the result, and you can revoke the connection at any time.
- The command line: Run publidock publish ./dist from any terminal, with a token in PUBLIDOCK_TOKEN.
- A GitHub Action: publidock/publish-action publishes from your workflow on every push you choose.
- An HTTP API with scoped tokens: A token can reach the whole workspace or only the sites you pick. Its value is shown once.
Guides
- How to let Claude, ChatGPT or Cursor publish to a link
- How to share a Claude artifact with people who don't use Claude
- How to turn a ChatGPT canvas into a shareable web page
- Share a vibe-coded prototype with your team
- The assistant that built it can ship it
Which AI tools work with Publidock?
Claude, ChatGPT, Cursor, Claude Code, Codex, VS Code, Gemini CLI, and any client that speaks MCP.
Do I have to paste an API key into my assistant?
No. You add Publidock as a connector and sign in when asked. You can revoke the connection at any time.
Is this on every plan?
Yes. The MCP server, the HTTP API with scoped tokens, the CLI and the GitHub Action are on every plan.