agentleFS
Boxline-devBoxline-dev/examples/skills/web/SKILL.md

web

Skill1 starsChanged 35 days agoMITView on GitHub

Map a live website into web.md, a guide coding agents can follow - every page, link, button, form, flow and auth wall, from the pages as a real browser renders them (Boxline). Use when asked to map, document or "learn" a website, or before automating one. Usage /web <url>.

  • Reads credentials

12 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write web.md: what a public website has and how to get around it, from its rendered pages.

    line 8
  2. Distinctive:

    Use whichever is there:

    line 12
  3. Distinctive:

    Only what was found: no guessed pages, buttons or flows.

    line 78
  4. Distinctive:

    No emojis. Keep it scannable.

    line 80

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/web: a site map for agents

Write web.md: what a public website has and how to get around it, from its rendered pages.

Boxline tools

Use whichever is there:

  • The Boxline MCP server: web_fetch (format html or markdown), web_search.
  • The boxline CLI (npx @boxline/cli …, with BOXLINE_API_KEY set or after npx @boxline/cli login).

Steps

  1. The address. Take the URL the user gave; add https:// when it has no scheme.
  2. The pages. Crawl the site (same host, robots.txt respected, at most 15 pages):

    npx @boxline/cli crawl <url> --limit 15 --format html -o .boxline-web

    With only the MCP server: web_fetch the start page as html, take its same-host links (no anchors, assets or duplicate addresses), and fetch up to 14 of them. A page that fails is skipped.

  3. Each page, from its HTML. Record only what is there:
    • navigation: main menu, sidebar, breadcrumbs, footer links (label → address);
    • actions: buttons (their exact label), links that act as buttons, dropdowns, toggles, what opens a dialog;
    • forms: each field's label, name, type, required, autocomplete, and the form's method and action;
    • flows: sign-up, sign-in, search, checkout, multi-step forms, as numbered steps;
    • data: tables, lists, repeated cards, filters, sorting, pagination;
    • walls: pages behind a sign-in or paywall, cookie banners, CAPTCHAs.
  4. The map. Routes and what each holds, how pages link, the main paths (home → pricing → sign-up), and the parts every page shares (header, footer).
  5. Write web.md in the project root in the format below, then delete .boxline-web.

Format

# web.md

> Site map and interaction guide for <url>, made with Boxline on <YYYY-MM-DD>

## Overview
- Site: <url>
- Pages mapped: <n>
- Interactive elements: <n>
- What it is: <e.g. marketing site with docs and a sign-up>

## Routes
| Route | Kind | What is there |
|---|---|---|

## Navigation
- Main menu: …
- Footer: …

## Pages
### <route>
**Actions:** "<exact label>" → <where it goes or what it does>
**Forms:** <field list, method, action>
**Sections:** …

## Flows
### Sign up
1. …

## Behind a sign-in
- <route>: <what the wall says>

Rules

  • Public pages only. Do not sign in, submit forms, solve CAPTCHAs or get around a wall: note it as behind a sign-in and move on.
  • Only what was found: no guessed pages, buttons or flows. Quote labels exactly.
  • Steps an agent can follow: addresses, labels and field names, not descriptions of the design.
  • No emojis. Keep it scannable.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.