Dataify MCP CLI

dataify-mcp-cli is a cross-platform command-line tool for calling tools exposed by the Dataify MCP HTTP service. After installation, use the dataify command on Windows, macOS, or Linux.

中文用户使用文档请查看 NPM_USAGE.md。

Requirements

• Node.js >= 18.17
• npm
• A valid Dataify API token

Check your local versions:

node -v
npm -v

Install

Install the public npm package globally:

npm install -g dataify-mcp-cli

After installation, the command is:

dataify

Verify the installation:

dataify --version
dataify --help

Configure API Token

Set your Dataify API token before calling tools:

dataify config set --token <your_api_token>

Example:

dataify config set --token YOUR_TOKEN

View the current config:

dataify config get

Show the config file path:

dataify config path

The config file is stored under your user home directory:

~/.dataify-mcp-cli/config.json

Token priority is:

--token -> DATAIFY_API_TOKEN -> saved config

Init

Run the setup wizard:

dataify init

Non-interactive examples:

dataify init --token YOUR_TOKEN --yes
dataify init --token YOUR_TOKEN --skip-mcp --skip-skill

The wizard uses an existing token when available; in an interactive terminal it can prompt for a token, save it, then install MCP configs and Dataify skills.

Quick Start

List the tools available to your token:

dataify tools

Show your account balance:

dataify balance

Show a tool's parameter names and descriptions:

dataify schema google_search

Choose a SERP, scraper, or web unlocker tool interactively:

dataify serp
dataify scraper
dataify webunlock

Install MCP configs for agent tools:

dataify mcp

Install Dataify skills:

dataify skill

Call a tool directly:

dataify google_search --q "pizza" --json 1

Or use the generic call form:

dataify call google_search --q "pizza" --json 1

Call the web unlocker:

dataify request_web_unlocker --url https://example.com --type html

Interactive Mode

Start interactive mode:

dataify

Or explicitly:

dataify chat
dataify repl

Interactive mode is a lightweight CLI loop. It does not connect to an AI assistant or interpret natural language; it lets you run Dataify MCP tool commands repeatedly.

Interactive commands can use a leading slash:

/help
/init
/tools
/balance
/serp
/scraper
/webunlock
/schema google_search
/call google_search --q "pizza" --arg-json json=1
/mcp
/skill
/retry
/clear
/exit

The slash is optional:

tools
balance
schema google_search
google_search --q "pizza" --arg-json json=1

Run With npx

You can run the CLI without a global install:

npx dataify-mcp-cli --help

Call a tool with npx:

npx dataify-mcp-cli google_search --q "pizza" --json 1 --token YOUR_TOKEN

If you use npx often, configure the token globally with dataify config set --token YOUR_TOKEN, or use the DATAIFY_API_TOKEN environment variable.

Category Wizards

The serp, scraper, and webunlock commands first list the tools in that category, then print the selected tool schema, then open one editable command line with defaults filled in.

Example:

dataify serp google_search --q "pizza" --json 1
dataify scraper amazon_product --url "https://www.amazon.com/dp/example"
dataify webunlock request_web_unlocker --url https://example.com --type html

Environment Variables

Environment variables override the saved config file.

Windows PowerShell:

$env:DATAIFY_API_TOKEN="YOUR_TOKEN"

Windows cmd.exe:

set DATAIFY_API_TOKEN=YOUR_TOKEN

macOS/Linux shells:

export DATAIFY_API_TOKEN="YOUR_TOKEN"

Optional request timeout:

Windows PowerShell:

$env:DATAIFY_MCP_TIMEOUT="3m"

Windows cmd.exe:

set DATAIFY_MCP_TIMEOUT=3m

macOS/Linux shells:

export DATAIFY_MCP_TIMEOUT="3m"

List Tools

View tools available to the current token:

dataify tools

Print the raw MCP response:

dataify tools --raw

Show Tool Schema

View the parameters for a tool:

dataify schema <tool_name>

Example:

dataify schema google_search

The output includes parameter names, required flags, types, and descriptions.

Call Tools

The basic form is:

dataify <tool_name> --param value

Example:

dataify google_search --q "pizza" --json 1

The generic form is equivalent:

dataify call google_search --q "pizza" --json 1

Unknown --name value flags are sent as MCP tool arguments. Dashes are converted to underscores, so --no-cache true becomes no_cache.

Argument Forms

Plain arguments are passed as strings:

dataify google_search --q "pizza" --json 1

This sends:

{
  "q": "pizza",
  "json": "1"
}

Use --arg-json for numbers, booleans, objects, arrays, or other JSON values:

dataify google_search --q "pizza" --arg-json json=1
dataify example_tool --arg-json enabled=true

Use a full JSON object in PowerShell, macOS, or Linux shells:

dataify google_search --args-json '{"q":"pizza","json":1}'

For cmd.exe, prefer --args-file to avoid JSON quote escaping issues.

Use a JSON file:

{
  "q": "pizza",
  "json": 1
}

dataify google_search --args-file params.json

Common Examples

Google Search:

dataify google_search --q "pizza" --arg-json json=1

Web Unlocker:

dataify request_web_unlocker --url https://example.com --type html

Output Options

Print the full MCP tool result:

dataify google_search --q "pizza" --json 1 --raw

Write output to a file:

dataify google_search --q "pizza" --arg-json json=1 --output result.json

Timeout

Some real-time collection tools can take longer to finish. Set a timeout with --timeout:

dataify google_search --q "pizza" --arg-json json=1 --timeout 3m

Supported timeout formats:

120000
30s
2m

Debug

Use --debug to print the request URL and JSON-RPC request body to stderr:

dataify google_search --q "pizza" --arg-json json=1 --debug

Debug output can include your token in the request URL. Do not share debug logs publicly.

Update

Update to the latest version:

npm install -g dataify-mcp-cli@latest

Check the installed version:

dataify --version

Uninstall

npm uninstall -g dataify-mcp-cli

Fixed MCP Endpoint

The CLI uses a fixed MCP endpoint:

https://mcp.dataify.com/mcp?token=<your_api_token>&tools=user_info,web_unlocker,google_serp,yandex_serp,duckduckgo_serp,bing_serp,amazon,youtube,facebook,instagram,reddit,walmart,google,booking,indeed,airbnb,google_play_store,github,tiktok,linkedin,glassdoor,twitter,crunchbase,zillow,ebay

Only <your_api_token> is configurable. The MCP server URL and tools query parameter are fixed in the CLI.

Local Development

Use these commands only when working from this repository checkout:

git clone https://github.com/dataify-server/cli.git
cd cli
npm install
npm install -g .

Run the local checkout without global installation:

npx . tools --token YOUR_TOKEN

Validate JavaScript syntax:

npm run check

Troubleshooting

dataify command not found

Close and reopen your terminal, then try:

dataify --help

If the command is still missing, check whether your npm global install directory is in PATH:

npm config get prefix

tools returns no tools

Check whether the token is configured correctly:

dataify config get

Set the token again if needed:

dataify config set --token YOUR_TOKEN

schema cannot find a tool

List the tools available to the current token:

dataify tools

If the tool is not listed, the current token or MCP service did not return that tool.

google_search returns Collection failed

google_search is a real-time collection tool. Target-site behavior, proxy/network conditions, or collection queues can affect a request.

Try a longer timeout:

dataify google_search --q "pizza" --arg-json json=1 --timeout 3m

Use debug output to inspect the request:

dataify google_search --q "pizza" --arg-json json=1 --debug

Token safety

Your token is stored in your local user config file and is sent to the Dataify MCP service as the token query parameter. Do not commit tokens to GitHub or share logs that contain tokens.