# MCP Server Setup

Connect real-time search to your AI client. AnySearch MCP Server provides web search, vertical search, parallel batch search, and URL content extraction through a hosted Streamable HTTP endpoint.

Hosted service, no local deployment

API key optional

4 available tools

### MCP SERVER · STREAMABLE HTTP

```
https://api.anysearch.com/mcp
```

### Web & vertical search

Search a question and get relevant results.

### Discover vertical domains

Look up the directory before setting parameters.

### Parallel batch search

Run 1–5 independent queries in parallel.

### Extract page content

Convert a web page to Markdown.

1. Choose access
2. Configure client
3. Verify setup

## Choose access

Try every tool anonymously, or configure an API key for higher request rates.

### Use an API key

Create a key in the console, then configure your client.

[create an API key in the console](/console/api-keys)

then add authentication as required by your client below. Most examples read the key from an environment variable. VS Code uses a password prompt; Antigravity uses private user settings.

### Set environment variable · current terminal session · Windows PowerShell

```powershell
$env:ANYSEARCH_API_KEY = "YOUR_API_KEY"
```

### Set environment variable · current terminal session · macOS / Linux

```bash
export ANYSEARCH_API_KEY="YOUR_API_KEY"
```

Replace YOUR_API_KEY with your key. This variable applies only to the current session. Start the client from the same terminal, or use a persistent environment setting supported by the client. Do not commit your key to a repository.

### Anonymous

No key required. Lower rate limits apply.

All client examples below now use anonymous access. No API key or environment variable is needed; keep X-Anysearch-Client. Anonymous access has lower rate limits. Configure an API key when you need more requests.

Your account or the actual response determines rates and quotas. Changing access updates every client configuration below.

## Choose your client and copy its configuration

These configurations use remote HTTP. Open the file for your client and merge the complete configuration. No SSE/stdio protocol conversion proxy is needed.

| Client | Configuration location | Connection | Setup |
| --- | --- | --- | --- |
| Cursor | .cursor/mcp.json | Direct HTTP | [View](/docs/mcp-install#config-cursor) |
| Claude Code | .mcp.json | Direct HTTP | [View](/docs/mcp-install#config-claude-code) |
| VS Code | .vscode/mcp.json | Direct HTTP | [View](/docs/mcp-install#config-vscode) |
| OpenCode | opencode.json | Direct HTTP | [View](/docs/mcp-install#config-opencode) |
| Cline | MCP Servers → Configure | Direct HTTP | [View](/docs/mcp-install#config-cline) |
| OpenAI Codex | \~/.codex/config.toml | Direct HTTP | [View](/docs/mcp-install#config-codex) |
| Google Antigravity | \~/.gemini/config/mcp\_config.json | Direct HTTP | [View](/docs/mcp-install#config-antigravity) |
| DeepSeek Harness | \~/.dsh/profiles/web/cordis.patch.yml | Direct HTTP | [View](/docs/mcp-install#config-deepseek) |
| Hermes Agent | $HERMES\_HOME/config.yaml | Direct HTTP | [View](/docs/mcp-install#config-hermes) |
| OMP (Oh My Pi) | .omp/mcp.json | Direct HTTP | [View](/docs/mcp-install#config-omp) |

Use a client version that supports remote HTTP. File locations and variable syntax vary by client; check each client’s official documentation for compatibility.

### Choose your client and copy its configuration

Choose your client below, then copy its configuration and follow the setup steps.

### Cursor

Configuration file: .cursor/mcp.json

User settings / other location: ~/.cursor/mcp.json

Direct HTTP · No proxy

#### With an API key

1. Create .cursor/mcp.json in the project root. To share the configuration across projects, use \~/.cursor/mcp.json.
2. Merge the configuration below into the file, keeping any existing MCP servers.
3. Reload the MCP configuration. Check Customize → MCPs or Output → MCP Logs for the connection.

Before starting Cursor, make sure its process can read ANYSEARCH_API_KEY. If you set it in a terminal, start the client from that terminal.

### JSON · Cursor

```json
{
  "mcpServers": {
    "anysearch": {
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "Authorization": "Bearer ${env:ANYSEARCH_API_KEY}",
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

#### Anonymous access

1. Create .cursor/mcp.json in the project root. To share the configuration across projects, use \~/.cursor/mcp.json.
2. Merge the configuration below into the file, keeping any existing MCP servers.
3. Reload the MCP configuration. Check Customize → MCPs or Output → MCP Logs for the connection.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · Cursor

```json
{
  "mcpServers": {
    "anysearch": {
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://cursor.com/docs/mcp)

#### Check connection status

### Verification · Cursor

```
agent mcp list
agent mcp list-tools anysearch
```

Run these commands if Cursor CLI is installed. In the IDE, check MCPs and MCP Logs instead.

### Claude Code

Configuration file: .mcp.json

User settings / other location: ~/.claude.json (managed through the CLI)

Direct HTTP · No proxy

#### With an API key

1. Create .mcp.json in the project root and merge the mcpServers configuration below.
2. Start Claude Code. Approve the project-level MCP connection when the client prompts you on first use.
3. Run /mcp in the session to inspect the anysearch status and tools.

Set ANYSEARCH_API_KEY before starting Claude Code. The client expands ${ANYSEARCH_API_KEY} in the configuration.

### JSON · Claude Code

```json
{
  "mcpServers": {
    "anysearch": {
      "type": "http",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "Authorization": "Bearer ${ANYSEARCH_API_KEY}",
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

#### Anonymous access

1. Create .mcp.json in the project root and merge the mcpServers configuration below.
2. Start Claude Code. Approve the project-level MCP connection when the client prompts you on first use.
3. Run /mcp in the session to inspect the anysearch status and tools.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · Claude Code

```json
{
  "mcpServers": {
    "anysearch": {
      "type": "http",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://code.claude.com/docs/en/mcp)

#### Check connection status

### Verification · Claude Code

```
claude mcp get anysearch
claude mcp list
```

These commands inspect the server. Use /mcp inside a session to check status and tools.

### VS Code

Configuration file: .vscode/mcp.json

User settings / other location: MCP: Open User Configuration

Direct HTTP · No proxy

#### With an API key

1. Create .vscode/mcp.json in the workspace. For user settings, run MCP: Open User Configuration.
2. Merge the configuration below. Enter your API key when prompted on first start.
3. Run MCP: List Servers, select anysearch → Start, and use Show Output to inspect logs.

This example uses a password input; no environment variable is needed. Enter your API key when the server first starts.

### JSON · VS Code

```json
{
  "inputs": [
    {
      "type": "promptString",
      "id": "anysearch-api-key",
      "description": "AnySearch API key",
      "password": true
    }
  ],
  "servers": {
    "anysearch": {
      "type": "http",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "Authorization": "Bearer ${input:anysearch-api-key}",
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

#### Anonymous access

1. Create .vscode/mcp.json in the workspace. For user settings, run MCP: Open User Configuration.
2. Merge the configuration below.
3. Run MCP: List Servers, select anysearch → Start, and use Show Output to inspect logs.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · VS Code

```json
{
  "servers": {
    "anysearch": {
      "type": "http",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://code.visualstudio.com/docs/agent-customization/mcp-servers)

#### Check connection status

### Verification · VS Code

```
MCP: List Servers
anysearch → Start / Show Output
```

These are VS Code Command Palette actions, not terminal commands. This example is for local VS Code; see the official documentation for Agent Host configuration.

### OpenCode

Configuration file: opencode.json

User settings / other location: ~/.config/opencode/opencode.json

Direct HTTP · No proxy

#### With an API key

1. Create opencode.json in the project root, or use \~/.config/opencode/opencode.json for user settings.
2. Merge the configuration below and keep oauth: false. API-key access does not require an OAuth login.
3. Restart OpenCode and run opencode mcp list in a terminal to check the connection.

Set ANYSEARCH_API_KEY before starting OpenCode. Its environment-variable syntax is {env:ANYSEARCH_API_KEY}.

### JSON · OpenCode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "anysearch": {
      "type": "remote",
      "url": "https://api.anysearch.com/mcp",
      "enabled": true,
      "oauth": false,
      "headers": {
        "Authorization": "Bearer {env:ANYSEARCH_API_KEY}",
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

#### Anonymous access

1. Create opencode.json in the project root, or use \~/.config/opencode/opencode.json for user settings.
2. Merge the configuration below and keep oauth: false. API-key access does not require an OAuth login.
3. Restart OpenCode and run opencode mcp list in a terminal to check the connection.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · OpenCode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "anysearch": {
      "type": "remote",
      "url": "https://api.anysearch.com/mcp",
      "enabled": true,
      "oauth": false,
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://opencode.ai/docs/mcp-servers/)

#### Check connection status

### Verification · OpenCode

```
opencode mcp list
```

Confirm that anysearch is connected. On Windows, the user configuration path is %USERPROFILE%\.config\opencode\opencode.json.

### Cline

Configuration file: MCP Servers → Configure

User settings / other location: Open the configuration file through the client

Direct HTTP · No proxy

#### With an API key

1. In Cline, open MCP Servers → Configure → Configure MCP Servers.
2. Merge the configuration and explicitly retain type: streamableHttp to avoid the old SSE mode.
3. Reconnect the server and check anysearch and its tools in the MCP panel.

Set ANYSEARCH_API_KEY before starting the IDE that runs Cline. The environment-variable syntax is ${env:ANYSEARCH_API_KEY}.

### JSON · Cline

```json
{
  "mcpServers": {
    "anysearch": {
      "type": "streamableHttp",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "Authorization": "Bearer ${env:ANYSEARCH_API_KEY}",
        "X-Anysearch-Client": "mcp/1.0.0"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}
```

#### Anonymous access

1. In Cline, open MCP Servers → Configure → Configure MCP Servers.
2. Merge the configuration and explicitly retain type: streamableHttp to avoid the old SSE mode.
3. Reconnect the server and check anysearch and its tools in the MCP panel.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · Cline

```json
{
  "mcpServers": {
    "anysearch": {
      "type": "streamableHttp",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://docs.cline.bot/mcp/mcp-overview)

#### Check connection status

### Verification · Cline

```
cline config mcp --json
```

CLI users can run cline mcp to open the interactive wizard. IDE configuration paths may vary by version; open the file through the menu.

### OpenAI Codex

Configuration file: ~/.codex/config.toml

User settings / other location: Trusted projects may also use .codex/config.toml

Direct HTTP · No proxy

#### With an API key

1. Open \~/.codex/config.toml. Trusted projects may also use .codex/config.toml.
2. Append the TOML configuration below while keeping your existing settings.
3. Reload the client. Run codex mcp list in a terminal, or /mcp in the TUI.

Set ANYSEARCH_API_KEY before starting Codex. bearer_token_env_var names the environment variable to read.

### TOML · OpenAI Codex

```toml
[mcp_servers.anysearch]
url = "https://api.anysearch.com/mcp"
bearer_token_env_var = "ANYSEARCH_API_KEY"
http_headers = { "X-Anysearch-Client" = "mcp/1.0.0" }
```

#### Anonymous access

1. Open \~/.codex/config.toml. Trusted projects may also use .codex/config.toml.
2. Append the TOML configuration below while keeping your existing settings.
3. Reload the client. Run codex mcp list in a terminal, or /mcp in the TUI.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### TOML · OpenAI Codex

```toml
[mcp_servers.anysearch]
url = "https://api.anysearch.com/mcp"
http_headers = { "X-Anysearch-Client" = "mcp/1.0.0" }
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://developers.openai.com/codex/mcp/)

#### Check connection status

### Verification · OpenAI Codex

```
codex mcp list
```

Use /mcp in the Codex TUI to inspect servers and tools.

### Google Antigravity

Configuration file: ~/.gemini/config/mcp_config.json

User settings / other location: Anonymous project configuration: .agents/mcp_config.json

Direct HTTP · No proxy

#### With an API key

1. Open the private user file \~/.gemini/config/mcp\_config.json. Anonymous mode can also use .agents/mcp\_config.json in the workspace.
2. Merge the configuration and retain serverUrl; do not replace it with url or httpUrl.
3. In the IDE, open MCP Servers → Manage MCP Servers, or run /mcp in the CLI.

Write the key only to private user settings in this example. The supplied reference does not specify environment interpolation for remote headers. Do not commit a plaintext key to project configuration.

### JSON · Google Antigravity

```json
{
  "mcpServers": {
    "anysearch": {
      "serverUrl": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0",
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

#### Anonymous access

1. Open the private user file \~/.gemini/config/mcp\_config.json. Anonymous mode can also use .agents/mcp\_config.json in the workspace.
2. Merge the configuration and retain serverUrl; do not replace it with url or httpUrl.
3. In the IDE, open MCP Servers → Manage MCP Servers, or run /mcp in the CLI.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · Google Antigravity

```json
{
  "mcpServers": {
    "anysearch": {
      "serverUrl": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://antigravity.google/docs/ide/mcp)

#### Check connection status

### Verification · Google Antigravity

```
MCP Servers → Manage MCP Servers
```

Check live status, logs, and tools in the client interface.

### DeepSeek Harness

Configuration file: ~/.dsh/profiles/web/cordis.patch.yml

User settings / other location: The corresponding profile under $DSH_HOME

Direct HTTP · No proxy

#### With an API key

1. Merge the patch into the web profile’s cordis.patch.yml, or the corresponding profile under $DSH\_HOME.
2. Run --dump-config to confirm that mcp-anysearch is present, then start the web profile.
3. Inspect the tools in the client and run a search. This MCP plugin exposes tools only.

Set ANYSEARCH_API_KEY before starting the web profile. The official MCP client plugin reads the environment variable.

### YAML · DeepSeek Harness

```yaml
- insert:
    - id: mcp-anysearch
      name: "@deepseek-ai/dsh-mcp-client"
      config:
        serverName: anysearch
        transport: streamable-http
        url: "https://api.anysearch.com/mcp"
        headers:
          Authorization: !!js '`Bearer ${process.env.ANYSEARCH_API_KEY}`'
          X-Anysearch-Client: "mcp/1.0.0"
        failOnStartupError: true
```

#### Anonymous access

1. Merge the patch into the web profile’s cordis.patch.yml, or the corresponding profile under $DSH\_HOME.
2. Run --dump-config to confirm that mcp-anysearch is present, then start the web profile.
3. Inspect the tools in the client and run a search. This MCP plugin exposes tools only.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### YAML · DeepSeek Harness

```yaml
- insert:
    - id: mcp-anysearch
      name: "@deepseek-ai/dsh-mcp-client"
      config:
        serverName: anysearch
        transport: streamable-http
        url: "https://api.anysearch.com/mcp"
        headers:
          X-Anysearch-Client: "mcp/1.0.0"
        failOnStartupError: true
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://github.com/deepseek-ai/deepseek-harness/blob/master/packages/mcp/mcp-client/README.md)

#### Check connection status

### Verification · DeepSeek Harness

```
npx @deepseek-ai/dsh --profile web --dump-config
npx @deepseek-ai/dsh web
```

DeepSeek Harness is in developer preview. Recheck the official plugin documentation when upgrading.

### Hermes Agent

Configuration file: $HERMES_HOME/config.yaml

User settings / other location: Defaults to ~/.hermes on macOS/Linux

Direct HTTP · No proxy

#### With an API key

1. Open $HERMES\_HOME/config.yaml and merge the mcp\_servers configuration below.
2. Restart Hermes. After changing configuration in a session, run /reload-mcp or start a new session.
3. Run hermes mcp test anysearch and hermes mcp list to inspect the results.

Put ANYSEARCH_API_KEY in $HERMES_HOME/.env or the client process environment.

### YAML · Hermes Agent

```yaml
mcp_servers:
  anysearch:
    url: "https://api.anysearch.com/mcp"
    headers:
      Authorization: "Bearer ${ANYSEARCH_API_KEY}"
      X-Anysearch-Client: "mcp/1.0.0"
```

#### Anonymous access

1. Open $HERMES\_HOME/config.yaml and merge the mcp\_servers configuration below.
2. Restart Hermes. After changing configuration in a session, run /reload-mcp or start a new session.
3. Run hermes mcp test anysearch and hermes mcp list to inspect the results.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### YAML · Hermes Agent

```yaml
mcp_servers:
  anysearch:
    url: "https://api.anysearch.com/mcp"
    headers:
      X-Anysearch-Client: "mcp/1.0.0"
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/features/mcp.md)

#### Check connection status

### Verification · Hermes Agent

```
hermes mcp test anysearch
hermes mcp list
```

The native Windows installer defaults to %LOCALAPPDATA%\hermes.

### OMP (Oh My Pi)

Configuration file: .omp/mcp.json

User settings / other location: ~/.omp/agent/mcp.json

Direct HTTP · No proxy

#### With an API key

1. Create .omp/mcp.json in the project root, or use \~/.omp/agent/mcp.json for global settings.
2. Merge the mcpServers configuration below.
3. In an OMP session, run /mcp reload, /mcp test anysearch, and /mcp list.

Set ANYSEARCH_API_KEY before starting OMP. Its configuration expands ${ANYSEARCH_API_KEY}.

### JSON · OMP (Oh My Pi)

```json
{
  "$schema": "https://raw.githubusercontent.com/can1357/oh-my-pi/main/packages/coding-agent/src/config/mcp-schema.json",
  "mcpServers": {
    "anysearch": {
      "type": "http",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "Authorization": "Bearer ${ANYSEARCH_API_KEY}",
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

#### Anonymous access

1. Create .omp/mcp.json in the project root, or use \~/.omp/agent/mcp.json for global settings.
2. Merge the mcpServers configuration below.
3. In an OMP session, run /mcp reload, /mcp test anysearch, and /mcp list.

Anonymous mode: no API key or environment variable is required. Authentication is removed; the client identifier is retained.

### JSON · OMP (Oh My Pi)

```json
{
  "$schema": "https://raw.githubusercontent.com/can1357/oh-my-pi/main/packages/coding-agent/src/config/mcp-schema.json",
  "mcpServers": {
    "anysearch": {
      "type": "http",
      "url": "https://api.anysearch.com/mcp",
      "headers": {
        "X-Anysearch-Client": "mcp/1.0.0"
      }
    }
  }
}
```

Merge this configuration and keep your existing MCP servers.

[Client documentation](https://github.com/can1357/oh-my-pi/blob/main/docs/mcp-config.md)

#### Check connection status

### Verification · OMP (Oh My Pi)

```
/mcp reload
/mcp test anysearch
/mcp list
```

Run these commands inside an OMP session.

## Confirm the setup works

After saving the configuration, complete these 4 checks. Use the connection status, tool list, and search results shown by your client.

CHECK 01

### Server connected

Check that `anysearch` is connected in your client. See “Check connection status” above for client-specific commands.

CHECK 02

### 4 tools available

`search` `get_sub_domains` `batch_search` `extract`

CHECK 03

### First search completed

Send the prompt below. Confirm that the client calls AnySearch and inspect the actual search results.

CHECK 04

### Know where to troubleshoot

Check client logs for connection and authentication errors. Use the [FAQ](/docs/mcp-install#faq) to troubleshoot configuration and rate limits.

### Copy this prompt for your first search

```
Use AnySearch to search for “Model Context Protocol official documentation”. Return 3 results and keep the title and source URL for each result.
```

Expected: the client calls search and displays actual results. If no tool is called, check the MCP connection and tool permissions first.

## What do the 4 tools do?

Your client uses these MCP tools to search and extract content. Find their purposes, parameters, and limits below.

| Tool | When to use it |
| --- | --- |
| [search](/docs/mcp-install#tool-search) | Search a question and get relevant results. |
| [get\_sub\_domains](/docs/mcp-install#tool-get_sub_domains) | Look up the directory before setting parameters. |
| [batch\_search](/docs/mcp-install#tool-batch_search) | Run 1–5 independent queries in parallel. |
| [extract](/docs/mcp-install#tool-extract) | Convert a web page to Markdown. |

### Web & vertical search (search)

For general search, provide a query. Before searching a specific domain, use get_sub_domains to discover valid domains, subdomains, and parameter definitions.

| Parameter | Type | Required | Description |
| --- | --- | --- | --- |
| query | string | Required | One natural-language query with a single intent. |
| domain | string | Optional | A domain such as finance, academic, or security; use values returned by get_sub_domains. |
| sub_domain | string | Optional | A subdomain routing key returned by the directory. |
| sub_domain_params | object | Optional | Use the parameter definitions returned by the directory; do not guess values. |
| max_results | integer | Optional | 1–10; defaults to 10. |

### Example tool arguments · search

```json
{
  "query": "Model Context Protocol official documentation",
  "max_results": 3
}
```

Use one intent per search. For independent questions, use batch_search.

### Discover vertical domains (get_sub_domains)

Returns a Markdown table of subdomains, descriptions, and parameter definitions. Call this tool before any search that specifies a domain.

| Parameter | Type | Required | Description |
| --- | --- | --- | --- |
| domain | string | Choose one | Look up one domain. |
| domains | string[] | Choose one | Look up up to 5 domains in one request. |

### Example tool arguments · get_sub_domains

```json
{
  "domains": [
    "finance",
    "academic"
  ]
}
```

Use either domain or domains. Follow the field names and parameter values returned by the directory.

### Parallel batch search (batch_search)

Each query object uses the same fields as search. A failure in one query does not block the others.

| Parameter | Type | Required | Description |
| --- | --- | --- | --- |
| queries | object[] | Required | 1–5 query objects, using the same fields as search. |

### Example tool arguments · batch_search

```json
{
  "queries": [
    {
      "query": "MCP official getting started guide",
      "max_results": 3
    },
    {
      "query": "Cursor MCP configuration documentation",
      "max_results": 3
    }
  ]
}
```

At most 5 queries per call. Each query should express a single intent.

### Extract page content (extract)

Provide an http:// or https:// webpage URL to retrieve its content. Use it to read a source page after searching.

| Parameter | Type | Required | Description |
| --- | --- | --- | --- |
| url | string | Required | The full http:// or https:// URL of an HTML webpage. |

### Example tool arguments · extract

```json
{
  "url": "https://example.com"
}
```

Only HTML pages are supported. Content longer than 50,000 characters is truncated.

For vertical search, call get_sub_domains → read the parameter definitions → call search. Do not guess subdomains or parameter values.

## FAQ

### Do I need an API key to get started?

No. Anonymous access supports every tool, with lower rate limits. For higher rates, create an API key in the console and use the authenticated configuration. Your account or the actual response determines the available quota.

### Do Cursor, VS Code, or Cline need a proxy?

This page uses client configurations that support remote HTTP and connect directly to https://api.anysearch.com/mcp. Cline must explicitly use type: streamableHttp; do not reuse an old SSE configuration. “No proxy” means no SSE/stdio protocol conversion process is needed; it does not guarantee network access from your machine.

### The configuration is saved. Why can’t I connect?

Confirm that the URL is complete and uses HTTPS, and that your client supports remote HTTP. Check that the configuration is in the active project or user settings. With an API key, also confirm that the client process can read the environment variable. Review the client logs for network or configuration errors.

### How do I fix an authentication error?

Check that the key is valid and Authorization uses Bearer, a space, and the key. Set environment variables before starting the client, then restart it; setting a variable in a different terminal is not enough. The VS Code example reads the key from a password prompt and does not need an environment variable.

### What should I do if I hit a rate limit?

Stop repeated retries and wait as instructed by the response. Follow Retry-After or the wait time in the error message when provided. Anonymous users can configure an API key; existing key users should check quota and usage in the console.

### Can extract read PDFs or other files?

extract supports HTML webpages and returns Markdown. Content exceeding 50,000 characters is truncated. It does not provide PDF or general-purpose file parsing.

### What if my client is not listed?

Check that the client supports remote HTTP MCP and follow its official documentation for the endpoint and authentication. Configuration fields and environment-variable syntax differ across clients, so do not directly copy another client’s JSON. Upgrade older versions or check their official compatibility notes.

<!-- docs-navigation -->

## Documentation navigation

- [Source](https://anysearch.com/docs/mcp-install)
- Index: [Docs index](https://anysearch.com/docs.md)
- Previous: [/v1/extract](https://anysearch.com/docs.md/api-endpoints/v1-extract)
- Next: [Skill Installation](https://anysearch.com/docs.md/skill-install)