CLI | Stophy

Give your agents the live web from your terminal. Try it with no key:

curl -fsSL https://stophy.dev/install | bash
stophy search "bun runtime"

Results print as rows, or as JSON with --json.

Install

The installer puts the binary in ~/.local/bin. On Windows, use PowerShell:

irm https://stophy.dev/install.ps1 | iex

With Node.js 20 or later, npm install -g @stophy/cli works too.

Set up your agent

stophy init --all

init logs you in with your browser, installs the Stophy skills with npx -y skills add stophydotdev/stophy, and adds the MCP server. With Claude Code on your PATH it runs claude mcp add. For any other agent it prints the command or snippet to use, and leaves your config files alone.

Option	What it does
`--all`	Run every step without asking
`--agent <name>`	Set up one agent, such as `claude-code` or `codex`
`--skip-auth`	Do not log in
`--skip-skills`	Do not install the skills
`--skip-mcp`	Do not add the MCP server

Log in

stophy login --browser

The CLI prints a code and opens stophy.dev. Check that the page shows the same code, then click Approve. On a server with no browser, open the printed link on any other device. To use an API key from the dashboard instead, run this and paste the key when it asks, so it stays out of your shell history. In CI, set STOPHY_API_KEY from your secret store.

stophy login --api-key

Each computer gets its own key, and logging in again replaces it. stophy logout logs out and revokes the key. Without a login you can run the commands stophy endpoints marks free, within a small free allowance.

Get data

stophy search "bun runtime" --limit 5
stophy scrape https://bun.com/docs
stophy map https://bun.com --limit 50
stophy crawl https://bun.com/docs --limit 20
stophy crawl status <id>
stophy extract --urls https://bun.com --prompt "the latest version"
stophy extract status <id>

stophy <command> --help lists every option.

Give a list as --urls a,b or --urls a --urls b.
Turn a yes-or-no option on with --flag and off with --no-flag.
Options that take JSON, like --formats '["markdown","links"]', take it as one quoted argument.
A value the command doesn’t take stops the CLI before it calls Stophy, and it prints the values that work.

crawl, batch scrape and extract answer with an id. Read the result with status <id> and stop a running job with cancel <id>. A status answer holds up to 100 finished pages. Wait until the status is completed. When it has a next link, run it again with --cursor and the cursor from that link.

Read a site with a parser

A parser returns a site’s own data as JSON: posts, videos, places, products, jobs, listings and more. stophy parsers [word] lists them, with the kind of page each one takes:

stophy parsers reddit
stophy scrape "https://www.reddit.com/r/rust/" --parser reddit-subreddit
stophy scrape https://youtu.be/M4TufsFlv_o --parser youtube-transcript

If the output ends with a cursor, get the next page with --parserOptions '{"cursor":"<cursor>"}'. --json prints the data as JSON and the cost to stderr. --raw prints the full response. -o <path> writes the output to a file. A failed call ends with Request id: <id>, which support will ask for.

Commands

Command	What it does
`stophy init`	Set up Stophy in your coding agents
`stophy login`	Log in with your browser or an API key
`stophy logout`	Log out and revoke this computer’s key
`stophy <command>`	Get data, for example `stophy search`
`stophy parsers [word]`	List the parsers `scrape --parser` takes
`stophy endpoints [word]`	List every command and its price
`stophy status`	Show your login, balance and CLI version
`stophy usage`	Show your balance and all-time usage
`stophy logs --days 7`	Show your recent requests
`stophy doctor`	Check your install, login and connection
`stophy version`	Show the CLI version

The CLI tells you when a newer version is out. Set STOPHY_NO_UPDATE_CHECK=1 to turn that off.

​Install

​Set up your agent

​Log in

​Get data

​Read a site with a parser

​Commands