<?xml version="1.0" encoding="UTF-8"?>
<reference anchor="I-D.wolfe-faf-format" target="https://datatracker.ietf.org/doc/html/draft-wolfe-faf-format-02">
   <front>
      <title>FAF: YAML Media Types for AI Project Context and Persistent Memory</title>
      <author initials="J." surname="Wolfe" fullname="James Wolfe">
         <organization>FAF Foundation</organization>
      </author>
      <date month="June" day="28" year="2026" />
      <abstract>
	 <t>   This document specifies two YAML-based media types that together form
   a foundational AI-context layer (FAF): .faf for static project
   context (what an AI agent KNOWS about a project) and .fafm for
   persistent agent memory (what it REMEMBERS across sessions).

   .faf (application/vnd.faf+yaml, registered in the IANA vendor tree
   2025-10-30) carries static project context — architecture,
   conventions, dependencies, goals — read once and treated as
   canonical. .fafm (application/vnd.fafm+yaml, registered 2026-05-13)
   carries persistent agent memory — facts, preferences, accumulated
   knowledge — which mutates across sessions and supports multiple
   profiles (voice and knowledge).

   The two formats share YAML 1.2 as base syntax and MIT-licensed
   distribution, but differ in lifecycle: .faf is static-read-once and
   .fafm is mutating-persisted.  This document documents the existing
   vendor-tree registrations and defines the &quot;faf&quot; well-known URI, a
   discovery anchor that returns a .faf context document and may
   reference associated memory and agent resources.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-wolfe-faf-format-02" />
   
</reference>
