Atomic Content Protocol

What is an ACO?

An Atomic Content Object is a Markdown file with structured YAML frontmatter. Six required fields. Everything else optional. Readable in any text editor. Parseable by any LLM.

---
id: "550e8400-e29b-41d4-a716-446655440000"
acp_version: "0.2"
object_type: "aco"
source_type: "manual"
created: "2026-04-10T12:00:00Z"
author: { id: "user-1", name: "Your Name" }
title: "My First ACO"
tags: ["example"]
---

Your knowledge content here.

Portable

Every ACO is a valid Markdown file. Works in Obsidian, any text editor, any static site generator — no proprietary tooling required.

AI-Ready

Token counts, summaries, typed entities, and agent permissions — everything AI needs to make smart decisions without reading the full content.

Provenance Built-In

Every auto-generated field carries provenance: which model created it, when, at what confidence. You always know what is human and what is machine.

Open Standard

ACP is a protocol, not a product. Apache 2.0 licensed. Anyone can implement it.

Three ways to create an ACO

1. Write one by hand (zero dependencies)

Create any .md file with the 6 required frontmatter fields. That is a valid ACO.

2. Use the CLI

npx @acp/cli init ./my-vault
npx @acp/cli create --title "My First ACO"
npx @acp/cli enrich ./my-vault/my-first-aco.md

3. Use the Stacklist MCP (zero-setup enrichment)

Call enrich_content or enrich_url — pass in content, get back a fully enriched ACO with tags, summary, entities, and provenance.