// Use case · AI agents

Give your agent the live web, with a price on every call

For teams building agents that must answer from today’s web: one MCP endpoint or one API key, eleven tools, and each tool’s cost in the description the model reads.

  • MCP tools

    11

  • credit per scraped page

    1

  • credits per search page

    2

POST /api/v1/scrape 200 OK
{  "success": true,  "data": {    "markdown": "# Pricing\n\n…",    "metadata": {      "title": "Pricing - Example",      "statusCode": 200    }  },  "credits": 1}
"credits" in the response
The call behind the scrape_page tool: markdown an LLM can read, and the exact charge. This call debited 1 credit.

// Why

An agent is only as current as its last fetch

The problem

An agent only knows what it was trained on

  • Prices, changelogs and competitors change after the training cutoff. Without a fetch at call time, the agent answers from memory and sounds sure about stale facts.
  • Browsers, proxies, HTML-to-markdown and retries are infrastructure you maintain instead of the agent.
  • A scrape or a crawl is a spend decision the model makes alone. If the price is not in front of it, the price is not in the decision.

What Wayfern does

Eleven tools on one endpoint, each with its price

  • One MCP connection, or one API key for REST, gives search, scrape, extract, map, crawl, screenshots and site reports.
  • The nine Web Data tools state their credit cost in the description the model reads, and every result reports credits_charged.
  • MCP and REST call the same service on the same balance, so a workflow moves from the agent to a batch job at the same price.

// How it works

From token to tool call in four steps

No SDK to install and no second integration for REST: the same token and the same balance serve both.

  1. Step 1 of 4

    Mint a token

    Create a wf_… token with the mcp scope in the dashboard. Add the web scope to use the same token over REST.

    Dashboard → API tokens

  2. Step 2 of 4

    Point your client at the endpoint

    Paste the JSON config into your MCP client. Streamable HTTP, stateless: no handshake, no session header.

    https://api.wayfern.com/mcp

  3. Step 3 of 4

    The agent picks a tool

    search_web, scrape_page, extract_data, map_site, crawl_site and six more, each with its price in the description.

    tools/list → 11 tools

  4. Step 4 of 4

    Every result reports its cost

    Results carry credits_charged. The balance is checked again on every call.

    "credits_charged": 1

// Connect

Point your client at one endpoint

Drop the JSON into your MCP client and the whole tool list is there. Prefer plain HTTP? The other tabs run the same scrape through the REST API with the same token.

Connect, or call it directly
{
  "mcpServers": {
    "wayfern": {
      "type": "http",
      "url": "https://api.wayfern.com/mcp",
      "headers": {
        "Authorization": "Bearer ${WAYFERN_API_TOKEN}"
      }
    }
  }
}
  • Endpoint

    Streamable HTTP. POST carries every message.

    https://api.wayfern.com/mcp

  • Auth

    A wf_… token with the mcp scope. The same token serves REST with the web scope. An active plan or a positive credit balance is required.

  • Revision

    Stateless. The older session-based shape that current SDKs still speak is served on the same endpoint.

    2026-07-28

// Tools and prices

What your agent gets, and what each call costs

This table renders from the same rate config the backend charges from, so the two cannot disagree.

  • search_web

    Credits
    2 per search
    What it does
    Google, Bing and DuckDuckGo results with excerpts, ads, pagination and related queries. Metered per provider result page fetched.
  • scrape_page

    Credits
    1 per page
    What it does
    One page as clean markdown, HTML, links and metadata.
  • extract_data

    Credits
    1 per source page
    What it does
    Structured data from up to 100 URLs against a prompt and an optional JSON Schema. Pages already fetched are not billed again.
  • map_site

    Credits
    2 per 10 URLs
    What it does
    A site's URLs from its sitemaps and page links, optionally filtered by a search term. Billed on the URLs returned.
  • crawl_site

    Credits
    1 per successful page
    What it does
    Start a bounded multi-page crawl. Returns a job id at once and charges nothing up front.
  • crawl_status

    Credits
    Free
    What it does
    Poll a crawl and read a page of its results. Pages were billed as they were crawled.
  • capture_screenshot

    Credits
    5 per capture
    What it does
    A page image with cookie banners dismissed and tall pages captured end to end. Returns a download URL.
  • brand_profile

    Credits
    10 per profile
    What it does
    Identity, colours, logos, social links and company metadata for a domain.
  • site_styleguide

    Credits
    10 per site
    What it does
    Colours, typography, spacing, radii, shadows and the design tokens a site declares.
  • site_fonts

    Credits
    5 per site
    What it does
    Font families, where each is served from, the weights available and where they are used.
  • list_search_engines

    Credits
    Free
    What it does
    The live provider ids and labels that search_web accepts.
Add-onsAsking scrape_page or crawl_site for a screenshot, PDF or JSON output adds 4 credits once per page, however many of the three you request; a summary adds 2 per page. Map bills per started block of 10 returned URLs. Free to read: the wayfern://search-engines and wayfern://mcp/help resources and the research_with_wayfern prompt.

// What it costs

Three agent sessions, priced out

Worked from the rates above. Each box ends with how many such runs fit in the 1,520 monthly credits of the $19 Developer plan.

Answer one question, with sources

Search once, then read the five results worth reading.

Credits for: Answer one question, with sources
ItemCredits
1 × search_web, one provider page2
5 × scrape_page to markdown5

Total

7credits

about $0.09
at $0.0125 / credit

About 217 runs a month on Developer ($19 for 1,520 credits).

Brief the agent on a competitor

What the design and positioning of one domain can tell you.

Credits for: Brief the agent on a competitor
ItemCredits
1 × brand_profile10
1 × site_styleguide10
1 × site_fonts5
1 × capture_screenshot5

Total

30credits

about $0.38
at $0.0125 / credit

About 50 runs a month on Developer ($19 for 1,520 credits).

Pull a docs site into context

Discover the URLs, crawl them to markdown, poll until done.

Credits for: Pull a docs site into context
ItemCredits
map_site, 200 URLs returned40
crawl_site, 200 pages crawled200
crawl_status polling0

Total

240credits

about $3.00
at $0.0125 / credit

About 6 runs a month on Developer ($19 for 1,520 credits).

The same balance covers these tools, browser sessions, captcha solves and residential bandwidth. Every published rate is on the pricing page.

// FAQ

Questions agent builders ask

Which MCP clients work?

Any client that can connect to a remote Streamable HTTP server. Paste the JSON config with your token. The server speaks revision 2026-07-28 statelessly and also serves the older session-based shape that current SDKs still speak, on the same endpoint.

Does a tool call cost more than the same REST call?

No. Each MCP tool calls the same domain service its REST endpoint calls, so validation, behaviour and billing come from one place. A tool result carries credits_charged; a REST response carries credits in the same envelope.

How does the model know what a tool costs?

The nine Web Data tools state their credit cost in the description the model reads, filled in from the backend's rate config. The server's instructions add that every fetching tool spends credits and tell the model to prefer the cheapest tool that answers the question.

What happens when a site blocks the fetch?

scrape_page, extract_data, capture_screenshot and the three site reports take a proxy option that defaults to "auto": the run goes out on a datacenter exit, and a navigation classified as blocked is retried once through a residential exit. Datacenter egress is free; residential bandwidth is metered from the same balance. Set "datacenter" or "residential" to decide yourself.

What if the task needs a login or several steps?

The tools cover fetch-shaped work. To sign in, step through a flow or keep one identity across many actions, drive a browser session over CDP at wss://api.wayfern.com/ws with Playwright or Puppeteer, on the same account and credit balance.

What happens when the balance runs out?

Every request checks that the account still has an active plan or a positive balance. At zero, calls are refused, unless you turned on auto top-up (off by default), which bills the overage instead.

// Next step

Connect your agent in one config block

Mint a token, paste the config, and your agent has eleven tools and one balance. Every call reports what it cost.