Feed Type: manifesto.llmfeed.json

Purpose

A manifesto.llmfeed.json is a signed, structured declaration of intent, ethics, and principles, meant to be read by LLM agents.

It helps agents and users understand what a site stands for — beyond APIs and content.

Where to expose

Sites are encouraged to publish this file at:

.well-known/manifesto.llmfeed.json

And optionally reference it in their:

  • llm-index.llmfeed.json
  • certification processes
  • spec pages or footers

Core Structure

{
  "feed_type": "manifesto",
  "metadata": {
    "title": "LLMFeed Manifesto",
    "origin": "https://wellknownmcp.org",
    "description": "Declaration of intent and ethical foundation for MCP and LLMFeed",
    "lang": "en",
    "generated_at": "2025-05-19T14:00:00Z"
  },
  "intentions": [
    {
      "area": "decentralization",
      "value": "We aim to empower any site, not just platforms, to speak to agents."
    },
    {
      "area": "trust",
      "value": "We believe trust must be visible, verifiable, and revocable."
    },
    {
      "area": "governance",
      "value": "We encourage public participation and nonprofit certification."
    }
  ],
  "ethics": [
    {
      "principle": "Agent agency",
      "statement": "Users must be able to understand and control what agents can do."
    },
    {
      "principle": "Open access",
      "statement": "No central registry is required to publish a LLMFeed."
    }
  ],
  "trust": {
    "signed_blocks": ["feed_type", "metadata", "intentions", "ethics"],
    "scope": "partial",
    "certifier": "https://llmca.org",
    "public_key_hint": "https://wellknownmcp.org/.well-known/public.pem"
  }
}

Agent behavior recommendations

Scenario Expected action
Manifesto present Display to user in UI summary
Signed + certified Use for policy fallback in absence of prompt / capability info
Conflicting with other feeds Defer to signed feed, prompt for user review

Related pages

  • /tools/well-known
  • /about
  • /faq
  • /join
  • llmfeedhub/preview (may display manifesto summary)