Charming docs

Your AI builds the app; Charming hosts it, gives it a real URL, and keeps it running after the chat closes.

I’m new here

If you haven’t built anything yet, start with How Charming works.

To build your first app, paste this into your agent:

Read https://usecharming.com/start.md then help me create my first app.

start.md tells your agent whether to use MCP or the HTTP API, then walks it through the build.

Every Charming app can:

Remember data and files. Key-value storage for lists, settings, and totals, plus file storage for images, PDFs, and other assets your code reads back. Both survive code updates and outlive the chat. See Data storage.
Stay private. Apps are private by default: one owner, one dataset, with roles for anyone you invite. See Privacy and sharing.
Invite people. Share an app as a collaborator, an end-user, or a read-only viewer. See Privacy and sharing.
Connect services. Call external APIs with secrets your app never sees in plaintext. See Secrets.
Connect to any AI agent. Build in one agent, read and update from another, over MCP or HTTP. See Connect to any AI agent.
Own your source. Your app’s code is yours: read the full source anytime and take it anywhere. See Code export.
Remix. Open any remixable app to get your own independent copy of its current source. See Remixing.

The agent-facing build manual and reference material live at these URLs:

llms-full.txt: the full build manual (module shape, capabilities, HTTP API, error kinds, worked examples).
build-mcp.md / build-http.md: the MCP and HTTP starter guides.
docs.md: the MCP tool list and the window.buildy UI API, in plain markdown.
openapi.json: the machine-readable OpenAPI 3.1 contract.
clients.txt: per-client setup strings for adding Charming as an MCP server.
pricing.md: free vs. claimed-account tiers, per-app size and asset caps, per-route rate limits, and token lifetimes.