Paperclip + Wayfern

Each agent in the company gets its own token.

  • One token per agent
  • Through the agent's adapter
  • Revoke one at a time

Paperclip runs a company of agents: a researcher, a support agent, a marketer, each with a role, a budget and a heartbeat. Many of their tasks start with a page on the live web.

Give each role its own Wayfern token through its adapter. The agent then scrapes, searches and extracts data, and you can revoke one agent's token and leave the others as they are.

A board with three columns: To do, Doing and Done. Cards for the research, support and marketing agents move across, each with its own Wayfern token chip, until all three are done.

Setup

Connect through the agent's adapter.

Paperclip documents MCP for the Claude Code and Hermes adapters. Other adapters use their own CLI's MCP settings.

  1. 1

    Install Paperclip

    Skip this when it runs. It needs Node 24.11 or later.

    bash

    npx paperclipai@latest onboard --yes
    npx paperclipai run
  2. 2

    Make a token per agent

    On the Tokens page, create one token with the mcp scope for each agent, named after its role.

  3. 3

    Add Wayfern to the adapter

    For a Claude Code adapter, run this as the OS user that runs the heartbeat.

    bash

    claude mcp add --transport http --scope user wayfern https://api.wayfern.com/mcp \
      --header "Authorization: Bearer $WAYFERN_API_TOKEN"
  4. 4

    Give the token as a secret

    Put the token in the adapter's env. Paperclip adds the value when the agent runs.

    json

    "env": {
      "WAYFERN_API_TOKEN": { "type": "secret_ref", "secretId": "...", "version": "latest" }
    }

Other ways to connect

  • Paperclip Connectors

    Open Connectors, Connect your own MCP server, and paste the URL. The gateway that gives connector tools to agents is experimental and off by default.

    URL

    https://api.wayfern.com/mcp
  • As an agent skill

    For a Claude Code adapter, put the skill in the heartbeat user's ~/.claude/skills. It reads WAYFERN_API_TOKEN from the adapter's env.

    bash

    mkdir -p ~/.claude/skills/wayfern
    curl -fsSL https://wayfern.com/skills/wayfern/SKILL.md -o ~/.claude/skills/wayfern/SKILL.md

Route

The adapter carries the tools.

Paperclip documents MCP through the Claude Code and Hermes adapters. A heartbeat runs with no person at a browser, so each agent uses its own token.

Remote MCP OAuth sign-in
In part
Paperclip Connectors take the URL. Their gateway to agents is experimental and off by default.
MCP + token Authorization header
Best fit
In the adapter's MCP settings, with one token per agent.
Agent skill SKILL.md
Works
A Claude Code adapter loads SKILL.md from ~/.claude/skills.
REST API api.wayfern.com/api/v1
Works
A process adapter calls api.wayfern.com/api/v1 with its token.

Example tasks

Work for each role

  • 01

    Research: build a lead list

    “Extract the name, website and city of the 40 exhibitors in the cold chain category from the trade fair's public exhibitor list. Attach a CSV to the issue.”

    extract_data bills 1 credit for each uncached source page.

  • 02

    Support: draft an outage reply

    “Read the vendor's public status page and changelog, and add a reply draft about the outage to the ticket.”

    scrape_page at 1 credit a page.

  • 03

    Marketing: check rankings

    “Search Google and Bing for our five target keywords and post where we rank to the CEO agent.”

    search_web bills 2 credits for each provider page.

  • 04

    Design: brand check

    “Get the brand profile of the new client's domain. Attach its colours, logos and social links to the onboarding issue.”

    brand_profile costs 10 credits for a domain.

Tools

What Paperclip 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

Budgets
Paperclip pauses an agent at 100% of its monthly budget. All tokens of one Wayfern account share one credit balance; get_usage shows it.
Evidence
Each Wayfern tool result has credits_charged. Put it in the issue with the result.