# Outcrawl documentation

> Outcrawl is web infrastructure for AI agents: one account, one credit balance, five capabilities: scrape, crawl, search, browsers, agents.

## Welcome
- /welcome/home: Platform overview
- /getting-started/quickstart: Make your first API call in under five minutes
- /getting-started/concepts: Credits, jobs, profiles, sessions and the shape every capability shares

## SDKs
- /sdks/overview: TypeScript vs Python, which to use
- /sdks/typescript: @outcrawl/sdk, zero dependencies, the only client with live browser control
- /sdks/python: outcrawl on PyPI, async and context managed

## CLI
- /cli/overview: A single static binary
- /cli/installation: curl and Homebrew on macOS and Linux, Scoop or a direct archive download on Windows
- /cli/agent: The full agent command family
- /cli/reference: Every command, generated from the registry

## API
- /api/overview: One base URL, one credential, two response framings
- /api/authentication: Bearer key authentication
- /api/errors: The closed 24-code error union

## Scrape
- /scrape/overview: One URL, clean structured content, 1 credit per page

## Crawl
- /crawl/overview: Walk a site from a seed, stream every page
- /crawl/jobs: Poll status, read results, cancel by id

## Search
- /search/overview: Web search, optionally rendered, 5 credits per 10 results

## Browsers
- /browsers/overview: Live CDP control, TypeScript SDK only
- /browsers/human-input: The measured human input model and why it matters
- /browsers/sessions: Replay and export every browser-driven run

## Agents
- /agents/overview: Hand over a task, billed for model usage, browser time and proxy usage (about 15 credits per minute, estimated)
- /agents/monitoring: Status, events, results, and agent-run webhooks (retried, unlike monitors)
- /agents/human-in-the-loop: Take the wheel of a live run, and hand it back
- /agents/files: Attach a file to a run

## Monitors
- /monitors/overview: Watch a URL on a schedule, notifications are not retried

## Account
- /account/authentication: Console sign-in
- /account/api-keys: The one credential every client reads
- /account/profiles: Saved logins and a pinned exit, kept across sessions
- /account/secrets: Credentials by handle, never read back
- /account/rules: Plain-language restrictions, enforced or guidance
- /account/integrations: Connect an external MCP server for an agent to call
- /account/settings: General workspace settings

## Billing
- /billing/overview: One prepaid credit balance
- /billing/pricing: Plans and rates for every metered kind
- /billing/credits: What a credit is, granted balance vs plan allowance
- /billing/usage: The aggregate of what every call already told you it cost
- /billing/payment-methods: Add a card from the console

## FAQ
- /faq/general: What Outcrawl is, who it is for, how to get support
- /faq/billing: How billing and the agent rate work
- /faq/capabilities: What each capability does and does not do
- /faq/security: Vulnerability reporting and credential handling

## MCP
- /mcp/overview: 27 tools, one per capability that carries one
- /mcp/setup: One command per client, sign in over OAuth, API key fallback for clients without it
- /mcp/tools: The full tool reference

## API reference
- /api-reference: Generated from the 46-row capability registry, never hand-written

## Examples
- /examples/cancel-forgotten-subscriptions: One agent per service walks the cancel flow and stops for your approval
- /examples/book-a-table-when-it-opens: A monitor spots an opening and an agent books it within your limits
- /examples/download-and-file-vendor-invoices: Agents sign in to each billing portal and file every invoice as a CSV
- /examples/tailored-job-applications: Draft tailored answers, approve them, then a second run submits
- /examples/price-a-basket-in-20-countries: Twenty agents, one per country, report what locals pay
- /examples/research-companies-into-your-crm: Search, crawl and agents profile 500 companies into HubSpot
- /examples/every-open-role-across-300-startups: Read 300 careers pages nightly and list what is new
- /examples/qa-your-signup-flow-hourly: Run your real signup flow every hour from three countries
- /examples/competitor-pricing-brief: A monitor catches a pricing change and an agent posts a brief to Slack
- /examples/appointment-slot-hunter: Book the first appointment slot that fits your week
- /examples/give-your-ai-app-a-browser: One browse(task) tool for your own assistant
- /examples/turn-a-website-into-an-api: Serve a signed-in portal with no API as typed JSON
- /examples/script-then-hand-to-the-agent: Playwright for the fixed steps, an agent for the rest
- /examples/docs-knowledge-base: Crawl a docs site into embeddings and re-embed only what changed
- /examples/deep-research-report: Search, read in parallel, and write a report where every claim has a URL
- /examples/mcp-monitor-competitor: Set up a page watch from your editor
- /examples/mcp-research-assistant: Search and scrape together in a conversation
- /examples/mcp-scrape-into-context: Pull a page into your coding assistant's context
- /examples/mcp-lead-extraction: Crawl a directory site from an assistant
- /examples/mcp-agent-from-editor: Submit and check on a long-running agent task
- /examples/mcp-connect-integration: Connect an integration, then let an agent use it