> ## Documentation Index
> Fetch the complete documentation index at: https://www.1password.dev/llms.txt
> Use this file to discover all available pages before exploring further.
# Search and read 1Password developer documentation with AI
> Search and read 1Password developer documentation from AI tools using a curated `llms.txt` index, per-page Markdown, `llms-full.txt`, and an MCP search server.
These options read documentation only. To give agents access to secrets or integrate with 1Password APIs, see [Secure AI access](/get-started/secure-ai-access) and [Build integrations](/get-started/build-integrations).
## Quickstart
Give your LLM one of these URLs depending on what you need:
| What you need | URL |
| ------------------------------------------------------ | ----------------------------------------------------------- |
| **Curated list** of key articles with descriptions | [`/llms.txt`](https://www.1password.dev/llms.txt) |
| **All documentation** in a single file | [`/llms-full.txt`](https://www.1password.dev/llms-full.txt) |
| **Any single page** as Markdown | Append `.md` to the URL of any article |
| **MCP server** to search and retrieve docs in AI tools | [`/mcp`](https://www.1password.dev/mcp) |
## Choose how you use AI
Choose the setup that matches your AI tool.
Use ChatGPT, Claude, Perplexity, or similar web chat to ask questions about 1Password developer tools.
Connect Claude Code, Cursor, Windsurf, VS Code, or another coding assistant so it can search docs while you write code.
Fetch documentation over HTTP for custom RAG pipelines, embedding indexes, or agent frameworks you run yourself.
## Ask in a chatbot
In web chat, share a documentation URL or paste content from an article.
### Ask about one page
On any article, select the down arrow next to "Copy page", then:
* Select **Open in ChatGPT**, **Open in Claude**, or **Open in Perplexity** to ask questions about that page.
* Copy the article as Markdown and paste it into your conversation.
You can also share the article's Markdown URL by appending `.md` to the page URL. For example, [`https://www.1password.dev/cli/get-started.md`](https://www.1password.dev/cli/get-started.md).
### Ask about the docs site
Share a URL and ask the model to read it:
* [`https://www.1password.dev/llms.txt`](https://www.1password.dev/llms.txt): Curated index of quickstarts, workflow guides, and key entry points. A good default when you want the assistant to orient itself, then fetch linked pages as needed.
* [`https://www.1password.dev/llms-full.txt`](https://www.1password.dev/llms-full.txt): All published documentation in one file. Use when you need exhaustive coverage in a single request and have a large enough context window.
## Code with an AI agent or IDE
This section covers AI tools that help you write code, not standalone web chat. Setup depends on your tool:
* **Terminal agent:** Run an AI from your terminal, like [Claude Code ](https://code.claude.com/docs/en/overview). It edits files, runs commands, and answers questions from the shell.
* **AI-native editor:** Use built-in AI chat or agent mode in your editor, like [Cursor ](https://cursor.com/home) Agent or [Windsurf ](https://windsurf.com/) Cascade.
* **IDE extension:** Add AI to a traditional editor, like VS Code with GitHub Copilot or [Claude ](https://claude.ai/).
**With MCP:**
Connect to the public documentation search server at `https://www.1password.dev/mcp` using [Model Context Protocol (MCP) ](https://modelcontextprotocol.io/). The server is public and requires no authentication. Your tool can search and retrieve documentation as Markdown during a session.
On any article, select **Copy page** → **Copy MCP Server** to copy the server URL.
Run this command in your terminal:
```shell theme={null}
claude mcp add --transport http 1password-dev https://www.1password.dev/mcp
```
**Without MCP:**
* Share [`/llms.txt`](https://www.1password.dev/llms.txt) in a Claude Code session and ask it to read the docs.
* For exhaustive coverage, share [`/llms-full.txt`](https://www.1password.dev/llms-full.txt).
* Select **Copy page** to copy the current article, or select **Open in Claude** to ask questions about the page.
**With MCP:**
Connect to the public documentation search server at `https://www.1password.dev/mcp` using [Model Context Protocol (MCP) ](https://modelcontextprotocol.io/). The server is public and requires no authentication. Your tool can search and retrieve documentation as Markdown during a session.
On any article, select **Copy page** → **Connect to Cursor** or **Copy MCP Server** to install the server or copy the URL.
In Cursor, add an entry to `.cursor/mcp.json`:
```json theme={null}
{
"mcpServers": {
"1password-dev": {
"url": "https://www.1password.dev/mcp"
}
}
}
```
**Without MCP:**
* In Cursor, use `@Docs` to add `https://www.1password.dev/llms.txt` as a documentation source.
* For exhaustive coverage, paste `https://www.1password.dev/llms-full.txt` in a chat or agent prompt.
* Select **Copy page** to copy the current article.
**With MCP:**
Connect to the public documentation search server at `https://www.1password.dev/mcp` using [Model Context Protocol (MCP) ](https://modelcontextprotocol.io/). The server is public and requires no authentication. Your tool can search and retrieve documentation as Markdown during a session.
On any article, select **Copy page** → **Copy MCP Server** to copy the server URL.
In Windsurf, add an entry to `~/.codeium/windsurf/mcp_config.json`. Open the file from **Windsurf Settings** → **Cascade** → **MCP Servers**, or see [Cascade MCP integration ](https://docs.windsurf.com/windsurf/cascade/mcp).
```json theme={null}
{
"mcpServers": {
"1password-dev": {
"serverUrl": "https://www.1password.dev/mcp"
}
}
}
```
**Without MCP:**
* Share `https://www.1password.dev/llms.txt` in a Cascade chat or agent prompt.
* For exhaustive coverage, paste `https://www.1password.dev/llms-full.txt` in a chat or agent prompt.
* Select **Copy page** to copy the current article.
**With MCP:**
Connect to the public documentation search server at `https://www.1password.dev/mcp` using [Model Context Protocol (MCP) ](https://modelcontextprotocol.io/). The server is public and requires no authentication. Your tool can search and retrieve documentation as Markdown during a session.
On any article, select **Copy page** → **Connect to VS Code** or **Copy MCP Server** to install the server or copy the URL.
Add to `.vscode/mcp.json`:
```json theme={null}
{
"servers": {
"1password-dev": {
"type": "http",
"url": "https://www.1password.dev/mcp"
}
}
}
```
See your extension's documentation if your setup uses a different config path.
**Without MCP:**
* Share `https://www.1password.dev/llms.txt` in a chat or agent prompt.
* For exhaustive coverage, paste `https://www.1password.dev/llms-full.txt` in a chat or agent prompt.
* Select **Copy page** to copy the current article.
## Build a custom agent or pipeline
If you're building an agent or automation pipeline and control how documentation is fetched, use HTTP retrieval or MCP.
### Direct HTTP retrieval
Use this when you manage retrieval yourself, such as custom RAG pipelines, embedding indexes, or batch jobs that don't use MCP.
* Start from [`/llms.txt`](https://www.1password.dev/llms.txt) for a curated map of entry points, then fetch linked pages as Markdown.
* Append `.md` to any article URL to fetch a specific page. For example, [https://www.1password.dev/cli/get-started.md](https://www.1password.dev/cli/get-started.md).
* Use [`/llms-full.txt`](https://www.1password.dev/llms-full.txt) when you need all published pages in one request.
### MCP
If your agent framework supports HTTP MCP servers, register `https://www.1password.dev/mcp` so the agent can search and retrieve documentation during a run.
Agents can also discover the server using [`/.well-known/mcp.json`](https://www.1password.dev/.well-known/mcp.json).
## Documentation formats
Not every format includes the same pages. [`llms.txt`](https://www.1password.dev/llms.txt) is a curated map of entry points; `llms-full.txt`, per-page `.md`, and MCP give you access to the full published documentation. [`llms.txt`](https://www.1password.dev/llms.txt) is the source of truth for what is in the curated map.
| Format | Coverage | URL | Best for |
| --------------- | --------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `llms.txt` | Curated entry points: quickstarts, workflow guides, and key overviews per product | [`https://www.1password.dev/llms.txt`](https://www.1password.dev/llms.txt) | Orienting an AI, then fetching linked pages for depth |
| `llms-full.txt` | All published pages (regenerated on every deploy) | [`https://www.1password.dev/llms-full.txt`](https://www.1password.dev/llms-full.txt) | Exhaustive coverage in a single request; requires a large context window |
| Per-page `.md` | One article | Append `.md` to any article URL | A specific topic. Also available from **Copy page** → **View as Markdown** |
| MCP | Search all published pages | [`https://www.1password.dev/mcp`](https://www.1password.dev/mcp) | Search and retrieve docs in AI tools and agents. See [Code with an AI agent or IDE](#code-with-an-ai-agent-or-ide) and [Build a custom agent or pipeline](#build-a-custom-agent-or-pipeline). Discoverable at [`/.well-known/mcp.json`](https://www.1password.dev/.well-known/mcp.json) |
## About the llms.txt standard
The [`llms.txt` standard](https://llmstxt.org/) is an open specification for making website content accessible to LLMs. It provides a structured, machine-readable Markdown index at a well-known URL (`/llms.txt`), so AI systems can discover and consume documentation without parsing HTML. The standard recommends curation over dumping an entire site into one file.