outcrawl

MCP

Connect once. Then ask in plain words.

MCP lets Claude Code, Codex or Cursor use outside tools. Add Outcrawl with one command and your agent reads and works the web for you, at 1 credit a page, about $0.0032 to $0.0049.

Connect your AI agentRead the MCP docs

1,000 free credits every month. No card needed.

Claude Codeoutcrawl
Pull in the authentication section of GitHub's REST API docs before we wire up the client. outcrawl_search: 3 results, 2.3 s, 1.5 credits. outcrawl_scrape: 178 lines of markdown, 0.9 s, 1 credit.

Real tool calls, captured 7 October 2026.

2.5 credits · $0.01225

Connect

One command, then sign in. No account or key needed first.

  1. 01

    Add Outcrawl.

    One command in Claude Code or Codex, or one line in Cursor. Nothing to install.

  2. 02

    Sign in when it asks.

    The first time your agent uses a tool, your browser opens our sign-in page. Signing in creates your account with 1,000 free credits.

  3. 03

    Ask.

    In plain words, the way you would ask a person. Your agent picks the tools.

terminalbash
claude mcp add --transport http outcrawl https://api.outcrawl.ai/mcp

Your browser opens to sign in the first time Claude Code calls a tool. No key to paste.

Any other MCP client that signs in with OAuth needs only the address: https://api.outcrawl.ai/mcp

Tools

27 tools your agent can use. It picks the right one for each request.

  • Scrape

    1 tool

    Read any page as clean markdown, or as JSON with the fields you name.

    • outcrawl_scrape

  • Crawl

    1 tool

    Read every page on a site from one link. All of them come back in one answer when the crawl ends.

    • outcrawl_crawl

  • Search

    1 tool

    Search the web, and read the top results in the same step when you ask.

    • outcrawl_search

  • Agents

    9 tools

    Hand off a task that needs clicks, forms or a sign-in. Check on it, answer what it asks, read what it found, take over, give it a file, or stop it.

    • outcrawl_agent
    • outcrawl_agent_list
    • outcrawl_agent_status
    • outcrawl_agent_events
    • outcrawl_agent_results
    • outcrawl_agent_cancel
    • outcrawl_agent_control
    • outcrawl_agent_answer
    • outcrawl_agent_file_add

  • Monitors

    3 tools

    Watch a page on a schedule and get a Slack, Discord or webhook message when it really changes.

    • outcrawl_monitor_create
    • outcrawl_monitor_list
    • outcrawl_monitor_delete

  • Saved logins

    4 tools

    Create, find and delete saved logins. A read or a run that uses one starts signed in.

    • outcrawl_profile_create
    • outcrawl_profile_list
    • outcrawl_profile_get
    • outcrawl_profile_delete

  • Session recordings

    4 tools

    Find a past run or session, open its recording, and see which requests failed.

    • outcrawl_sessions_list
    • outcrawl_sessions_get
    • outcrawl_session_network
    • outcrawl_session_request

  • Rules

    2 tools

    Read and set the rules every agent run must follow.

    • outcrawl_rules_get
    • outcrawl_rules_set

  • Usage and credits

    2 tools

    See what you have spent, on what, and what is left.

    • outcrawl_usage
    • outcrawl_credits

Examples

Ask the way you would ask a person. Real prompts from the docs, and the tools they call.

Ask your agentDocs

Set up a page watch without leaving your assistant.

example
Watch https://competitor.example/pricing every hour and post to our #competitors Slack channel if it changes. outcrawl_monitor_create: checking hourly, 1 credit a check. What am I currently watching? outcrawl_monitor_list: each monitor's page, schedule and last check.

Limits

Three things it leaves to you. Each one on purpose.

  • Passwords, cards and purchases

    Your agent reads web pages, and a page could ask it to. So a person stores passwords and cards from the CLI or the SDKs, and in the console connects the card providers runs pay with, reviews what runs bought, sets when a purchase waits for a yes, and connects tools such as Slack or Gmail.

    How agent runs use them

  • A live browser

    A live browser stays connected for the whole session, and a tool call answers once. Your agent writes a short Playwright script and runs it, or hands the task to an Outcrawl agent.

    How browsers work

  • A crawl in progress

    Your agent's crawl answers once, with every page, when the crawl ends. Checking on a crawl, paging through it and stopping it early are in the CLI and the SDKs.

    What the CLI does

Sign-in and safety

You sign in. Your agent never holds a key.

  • Sign in through your browser. Your agent’s first tool call opens our sign-in page. You approve it there and your client keeps the sign-in. No key is pasted anywhere.

  • Rules every run follows. Tell your agent “never spend over $50”, “only use amazon.com” or “never place an order”. It saves them with outcrawl_rules_set, and each comes back marked enforced, a gate that blocks the action, or guidance.

  • A ceiling on every agent run. A run started from MCP stops at 2,000 credits, about $6.39 to $9.80, unless your agent sets its own budget. It keeps what it found.

Pricing

Same credits, same rates. Billed per tool call.

Tool

Dollars on

  • Scrape

    Any web page as clean markdown or JSON.

    1 credit a page

    $4.90 per 1,000 pages

  • Search

    Web results for your agent. Reading a result page is 1 credit a page on top.

    5 credits per 10 results

    $2.45 per 1,000 results

  • Agents

    A task in plain English, done in a real browser.

    About 15 credits a minute

    about $0.07 a minute

Agent runs are billed for the AI model's usage, browser time and proxy usage, in credits. The estimate is a typical run; the actual cost depends on the task. The middle half of runs land between 10 and 25 credits a minute on Builder. Model usage converts at your plan's price per credit, so on Pro and Scale the same minute is more credits.

  • Your agent pays per tool call, at the same rates as the API, the CLI and the SDKs.

  • Every answer says what it cost, so your agent can tell you as it goes. outcrawl_usage and outcrawl_credits show the totals.

  • The session at the top of this page cost 2.5 credits, about $0.0080 to $0.0123.

See full pricing

FAQ

Questions about MCP