Skip to main content
These docs are built for two kinds of readers: people, and the AI tools people build with. This page covers both.
Two different MCP servers. The docs MCP on this page (docs.prava.space/mcp) lets an AI read these docs. The payments MCP (mcp.pay.prava.space/mcp) lets an agent actually pay; see Prava MCP.

​
Reading as a human

Start at the home page and pick the card that matches you. Each card opens a path you can read top to bottom: Stuck on a term? The Glossary maps every name across the API, CLI, and dashboard. Stuck on anything else? FAQ.

​
Reading as an AI tool or agent

Three ways, from lightest to deepest:
  1. Paste an index: llms.txt (page index) or llms-full.txt (entire docs as one file) into any LLM.
  2. Per-page: every page has a top-right AI menu (copy as Markdown, open in ChatGPT/Claude, view raw Markdown).
  3. Connect the docs MCP (below): your coding agent gets a live “search the Prava docs” tool and answers from current pages instead of stale training data.
And when the agent should act, not only read: connect the payments MCP too.

​
Connect the docs MCP server

We host a standard Model Context Protocol (MCP) server for these docs. MCP is an open protocol, so any MCP-compatible AI tool can connect to the same endpoint and gain a “search the Prava docs” capability: Claude Code, Cursor, VS Code / Copilot, Windsurf, Cline, Continue, Goose, Zed, and others. Add this to your client’s MCP settings:

​
Quick-add by tool

​
Older / stdio-only clients

If a client only supports local (stdio) MCP servers, bridge to the hosted endpoint with mcp-remote:
Once connected, ask your agent things like “Using the Prava docs, how do I create a payment session?” and it will pull the answer straight from these pages.

​
Feed the docs to any LLM (llms.txt)

Prefer to paste context into a model directly? These docs publish machine-readable indexes: Drop either URL (or its contents) into ChatGPT, Claude, or any LLM to ground it in Prava’s documentation.

​
Per-page AI actions

Every page has an AI menu in the top-right:
  • Copy page: copy the page as clean Markdown.
  • Open in ChatGPT / Claude: start a chat pre-loaded with this page.
  • View as Markdown: see the raw Markdown a model would read.

​
Why this matters for Prava

Prava is built for AI agents. Making the docs first-class for the tools your developers already use (Cursor, Claude Code, and others) means less guesswork, fewer hallucinated API calls, and a faster path from “what’s Prava?” to a working integration.