flyweb

Core types, validation, and helpers for the FlyWeb protocol — an open standard that makes the internet machine-readable.

robots.txt tells machines where not to go. flyweb.json tells them what you have.

Install

npm install flyweb

Usage

Define a config

import { defineConfig } from 'flyweb';

export default defineConfig({
  flyweb: '1.0',
  entity: 'TechCrunch',
  type: 'news',
  resources: {
    articles: {
      path: '/.flyweb/articles',
      format: 'jsonl',
      fields: ['title', 'author', 'date', 'summary', 'content', 'tags'],
      query: '?tag={tag}&limit={n}',
    },
    authors: {
      path: '/.flyweb/authors',
      format: 'json',
      fields: ['name', 'bio', 'avatar'],
    },
  },
});

Validate a config

import { validate } from 'flyweb';

const result = validate(someObject);

if (!result.valid) {
  console.error(result.errors);
  // ["resources.posts.path: must start with \"/\"",
  //  "resources.posts.format: must be one of json, jsonl, csv, xml"]
}

The Protocol

Every website publishes a file at /.well-known/flyweb.json that describes its structured data:

{
  "flyweb": "1.0",
  "entity": "TechCrunch",
  "type": "news",
  "resources": {
    "articles": {
      "path": "/.flyweb/articles",
      "format": "jsonl",
      "fields": ["title", "author", "date", "content"],
      "query": "?tag={tag}"
    }
  }
}

AI agents check this file first. No scraping. No guessing. Just structured data.

Types

import type { FlyWebConfig, FlyWebResource, EntityType, ResourceFormat } from 'flyweb';

| Type | Description | |------|-------------| | FlyWebConfig | Root config object served at /.well-known/flyweb.json | | FlyWebResource | A single resource (path, format, fields, query) | | EntityType | 'news' \| 'blog' \| 'ecommerce' \| 'saas' \| 'docs' \| ... | | ResourceFormat | 'json' \| 'jsonl' \| 'csv' \| 'xml' |

Framework Integrations

• Next.js — next-flyweb

Links

• flyweb.io — Project homepage
• GitHub — Source code
• X/Twitter — Updates

License

MIT