Kimi K3 + Wayfern

Long pages in, one clean file out.

~/.kimi-code/mcp.json
/mcp-config login
Skills in ~/.kimi-code/skills

Kimi Code runs K3 with a context of one million tokens, so long pages and big tables fit. Getting those pages from sites that block bots is the hard part.

Wayfern scrapes them, crawls whole sections and extracts tables in an anti-detect browser. Kimi reads the result, cleans it and writes the file.

~/coffee · kimi kimi-k3

› /mcp-config login wayfern

✓ wayfern connected · OAuth

› Extract the first 60 listings on

shop.acme.dev/pour-over into a CSV

Used extract_data (wayfern)

urls: …/pour-over?page=1, 2, 3

prompt: "title, price, rating, reviews"

60 items · credits_charged: 3

TitlePrice★Rev.

Ceramic cone dripper 02 $24.00 4.7 312

Glass server 600 ml $31.50 4.5 86

Gooseneck kettle 1 L $58.00 4.8 1,204

Paper filters, 100 pack $6.90 4.6 540

Hand burr grinder $74.00 4.4 97

… 55 more rows rows 01224364860

Used WriteFile (pour-over.csv)

pour-over.csv · 60 rows

Setup

Add Wayfern to Kimi Code.

Kimi Code runs on your computer and calls Wayfern over the remote server.

1
Install Kimi Code

Then start kimi and run /login. Homebrew and npm also work.
bash

curl -fsSL https://code.kimi.com/kimi-code/install.sh | bash
2
Add the server

Or run /mcp-config in kimi. An entry with url and no transport is Streamable HTTP.
~/.kimi-code/mcp.json

{ "mcpServers": { "wayfern": { "url": "https://api.wayfern.com/mcp" } } }
3
Sign in

Run this in kimi, then sign in on wayfern.com and click Allow. /mcp shows the status.
prompt

/mcp-config login wayfern

Ask

prompt

Call get_usage on Wayfern and tell me my credit balance.

Other ways to connect

With a token

Set WAYFERN_API_TOKEN to a token with the mcp scope. Kimi Code sends it as a Bearer token.

~/.kimi-code/mcp.json

{
  "mcpServers": {
    "wayfern": {
      "url": "https://api.wayfern.com/mcp",
      "bearerTokenEnvVar": "WAYFERN_API_TOKEN"
    }
  }
}

As an agent skill

Kimi Code loads skills from ~/.kimi-code/skills and ~/.agents/skills. The skill needs WAYFERN_API_TOKEN in the environment.
bash

mkdir -p ~/.kimi-code/skills/wayfern curl -fsSL https://wayfern.com/skills/wayfern/SKILL.md -o ~/.kimi-code/skills/wayfern/SKILL.md
kimi.com and Kimi Work

kimi.com has no custom MCP servers. Kimi Work's Plugin Builder makes a plugin from an MCP URL; its sign-in to Wayfern is not tested.

Route

Kimi reads, Wayfern fetches.

Kimi Code signs in to remote MCP servers with OAuth, and K3's context holds long tool results. A large crawl stays in Wayfern until Kimi reads it with crawl_status.

Remote MCP OAuth sign-in: Best fit; /mcp-config login wayfern.
MCP + token Authorization header: Works; bearerTokenEnvVar reads a wf_ token from a variable.
Agent skill SKILL.md: Works; Put SKILL.md in ~/.kimi-code/skills/wayfern. The Bash tool runs curl.
REST API api.wayfern.com/api/v1: Works; Kimi can write code that calls api.wayfern.com/api/v1 directly.
Browser API Playwright over CDP: Works; Kimi can write Playwright code that connects to a Wayfern browser over CDP.

Example tasks

Data jobs

01

Collect listings

“Extract title, price, rating and review count of the first 60 listings on this public shop search page. Save them as pour-over.csv.”

extract_data bills 1 credit for each uncached source page.
02

Crawl a catalogue

“Crawl the product section of example-supplier.com, 200 pages at most, and write one JSON row per product.”

crawl_site bills 1 credit a page. crawl_status reads the results for free.
03

Compare two languages

“Read the English and the German page of our 12 products and list each price or spec that differs.”

scrape_page at 1 credit a page: 24 credits for this job.
04

Count a site

“Map example.com and count its URLs by top-level folder.”

map_site bills 2 credits for each 10 URLs.

Tools

What Kimi K3 can call

The same tools on every route, at the REST API rates. Every result reports credits_charged.

All tool inputs

scrape_page Fetch one page through an anti-detect browser and return clean markdown, HTML, links and metadata. The screenshot, pdf and json formats add +4 credits (once); summary adds +2. 1 credit / page
extract_data Extract structured data from one or more pages with a prompt and an optional JSON Schema. Re-running over pages already fetched is free. 1 credit / uncached source page
map_site List a site's URLs from its sitemaps and page links. 2 credits / 10 URLs
capture_screenshot Capture a clean image of a page: cookie banners and ads removed, lazy images loaded, tall pages captured end to end. Returns a short-lived download URL and an artifact key, not image bytes. Slices of a full page cost nothing extra; a fail_if_contains match is not charged. 5 credits / capture
brand_profile Identity, colours, logos, social links and company metadata for a domain. 10 credits / profile
site_styleguide Colours, typography, spacing, radii, shadows and the design tokens a site declares. 10 credits / site
site_fonts Font families, where each is served from, available weights, file formats and where they are used. 5 credits / site
crawl_site Start a bounded multi-page crawl. Returns a job id immediately and charges nothing up front; pages are billed as the crawl runs. 1 credit / successful page
crawl_status Poll a crawl started with crawl_site and read a page of its results. The pages were billed as they were crawled. Free
search_web Search Google, Bing, and DuckDuckGo. Returns ranked results, ads, provider failures, pagination, and related queries. 2 credits / provider page
list_search_engines List live provider ids and labels. Free

Good to know

K3 and K3.1: Moonshot lists K3 as the previous generation and kimi-k3.1-pro as the current model. Kimi Code runs both.
Headless: kimi -p runs with auto permissions. Project MCP servers stay off until you trust the folder with KIMI_CODE_TRUST_WORKSPACE=1.
The old CLI: kimi mcp add exists only in the archived Python kimi-cli. Kimi Code uses mcp.json or /mcp-config.