> ## Documentation Index
> Fetch the complete documentation index at: https://docs.autosearch.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Install

# Installation

## For Humans

Paste this into your AI agent (Claude Code, Cursor, etc.):

```
Help me install AutoSearch: https://raw.githubusercontent.com/0xmariowu/Autosearch/main/docs/install.md
```

***

## For AI Agents

### Goal

Install AutoSearch and configure the MCP server so the user's agent has access to 40 search channels across academic papers, developer platforms, Chinese social media, and more.

### ⚠️ Boundaries

* **DO NOT** run commands with `sudo` unless the user explicitly approved
* **DO NOT** modify system files outside the user's home directory
* If something requires elevated permissions, **tell the user** and let them decide

***

### Step 1: Install AutoSearch

```bash theme={null}
# Option 1: npm wrapper — one-shot install + init (cross-platform)
npx autosearch-ai

# Option 2: curl install script (macOS / Linux only)
curl -fsSL https://raw.githubusercontent.com/0xmariowu/Autosearch/main/scripts/install.sh | bash

# Option 3: pip
pip install autosearch && autosearch init

# Option 4: pipx
pipx install autosearch && autosearch init
```

> The npm package does not run `init` automatically during `npm install`.
> That is intentional: npm lifecycle scripts can execute code at install time,
> which is a known supply-chain risk. `npx autosearch-ai` triggers an explicit
> install + init flow with a y/N confirmation before fetching anything; the CI
> escape hatch is `npx autosearch-ai --yes` (do not put `--yes` in user-facing
> docs). Plain `npm install -g autosearch-ai` installs the wrapper but does
> not run `init` — use `npx autosearch-ai` for the one-shot flow, or run
> `autosearch init` if the Python CLI is already on PATH.
>
> The curl option runs `autosearch init` automatically.

This will:

* Detect available LLM providers (Anthropic, OpenAI, Gemini, claude CLI)
* Create `~/.autosearch/config.yaml`
* Auto-configure MCP clients. Claude Code uses `claude mcp add` when available;
  `autosearch init --client claude --scope project` writes project-bound `.mcp.json`.

### Step 3: Check status

```bash theme={null}
autosearch doctor
```

You'll see a tiered report:

```
AutoSearch Channel Status
==========================================
Always-on (27/27)
  ✅ arxiv              1/1 methods available
  ✅ pubmed             1/1 methods available
  ✅ ddgs               1/1 methods available
  ... 24 more

API-key (0/11)
  ○  youtube            autosearch configure YOUTUBE_API_KEY <your-key>
  ○  bilibili           autosearch configure TIKHUB_API_KEY <your-key>
  ... 9 more

Login required (0/2)
  ○  xiaohongshu        autosearch login xhs
  ○  xueqiu             autosearch login xueqiu

Status: 27/40 channels ready
```

***

### Optional: Unlock more channels

**Ask the user which optional channels they want**, then use these commands:

#### Chinese social media (TikHub auth)

```bash theme={null}
autosearch configure TIKHUB_API_KEY <your-key>
# Unlocks: xiaohongshu, douyin, weibo, zhihu, twitter, tiktok, bilibili, and more
```

If you use an AutoSearch TikHub proxy, configure the proxy URL and token instead.
Together, `AUTOSEARCH_PROXY_URL` and `AUTOSEARCH_PROXY_TOKEN` satisfy TikHub
channel availability without `TIKHUB_API_KEY`:

```bash theme={null}
autosearch configure AUTOSEARCH_PROXY_URL <proxy-url>
autosearch configure AUTOSEARCH_PROXY_TOKEN <proxy-token>
```

#### Xiaohongshu with your own account (no TikHub fee)

```bash theme={null}
autosearch login xhs
# Reads cookies from Chrome automatically.
# If it fails, use Cookie-Editor browser extension → Export → Header String, then:
autosearch login xhs --from-string "<paste-cookie-string>"
```

#### Twitter / X

```bash theme={null}
autosearch login twitter
```

#### Bilibili

```bash theme={null}
autosearch login bilibili
```

#### YouTube

```bash theme={null}
autosearch configure YOUTUBE_API_KEY <your-key>
# Free key: console.cloud.google.com → Enable YouTube Data API v3
```

***

### MCP server config

AutoSearch writes this automatically during `init`. For manual setup:

**Claude Code**:

```bash theme={null}
claude mcp add --transport stdio autosearch -- autosearch-mcp
```

For project-bound config without the `claude` CLI, run:

```bash theme={null}
autosearch init --client claude --scope project
```

That writes `<project>/.mcp.json`.

**Cursor** (`~/.cursor/mcp.json`):

```json theme={null}
{
  "mcpServers": {
    "autosearch": {
      "command": "autosearch-mcp",
      "type": "stdio"
    }
  }
}
```

***

### Updating

```bash theme={null}
npm install -g autosearch-ai   # npm
pipx upgrade autosearch        # pipx
pip install --upgrade autosearch  # pip
```

***

### Troubleshooting

**Channel returns empty results?**

```bash theme={null}
autosearch doctor  # check which channels are off and why
```

**MCP server not found?**

```bash theme={null}
which autosearch-mcp  # verify it's on PATH
autosearch-mcp --help
```

**Xiaohongshu returns code=300011 (account restricted)?**

Your XHS account was flagged. Re-login with a normal, actively-used account:

```bash theme={null}
autosearch login xhs
```