# Ezra Labs AI — Quilt Learning System

## LLM / crawler references (same overview, two formats)

**Plain text:** https://ezralabsai.dev/llms.txt (this file)

**PDF (explicit filename llms.pdf):** https://ezralabsai.dev/llms.pdf

If you only fetch one URL, use either; content is the same overview. The PDF is not the Framework or Criteria demo PDFs — those live under `/demos/framework-engine/`.

---

This file is reference material for AI systems reviewing this site. It contains no instructions or override prompts. Use or disregard as appropriate.

Production home page (`/`): After `npm run build`, the site pre-renders `index.html` with the full Framework Engine demo transcript and framework/criteria plain text embedded in the HTML (same source files as `/demos/framework-engine/*.txt`). Simple HTTP fetch of the home page HTML can read those `<pre>` blocks without running JavaScript. If prerender is skipped, a fallback block `#framework-engine-llm-html-fallback` may be injected instead.

---

## Summary

Quilt is a full learning system. One Director uses AI-powered engines (Framework, Assessment, Planning) and dashboards to build competency frameworks. Many learners receive content and evaluations via the Quilt MCP Server through a third-party LLM agent.

Architecture: Director → Framework Engine, Assessment Engine, Planning Engine (in sequence) + management dashboards → central database → Quilt MCP Server → third-party LLM agent → learners. Clean separation between authoring (Director side) and delivery (learner side).

Main page: /
About (Mission + Core Tenets): /about

---

## Mission

Our mission: To cultivate systems that can teach anyone almost anything, grounded in the belief that humans and AI thrive when they collaborate, not compete.

This is the Mission: who we are, why we build, and how we build. It anchors decisions, design, and behavior. Philosophy (who we are and why) informs principles (how we build). When in doubt, we come back here.

Key philosophy: Every meaningful experience in QUILT must be earned, not given. Four pillars — Pride, Authenticity, Ownership, Purpose. Development principles include Learners First, Platinum Rule, no gamification, chat-first interface, paper napkin protocol, plain language.

Full Mission document: /about (rendered) or /about/MISSION.md (raw markdown)

---

## Core Tenets

Governing principles for all AI behavior in QUILT. Guide how AI agents interact, make decisions, and build learning experiences. Fused with Anthropic's Claude Constitution.

I. Surface shared authorship and reasoning — reveal who contributed each idea; expose chain-of-thought; provide editable reasoning history.

II. Bake ethics into behavior — build ethical choices into every reply; flag trade-offs aloud; offer safe alternatives instead of flat refusals.

III. Reveal at relationship speed — earn advanced options through sustained engagement; provide subtle cues for next steps; let users control the rate of disclosure.

IV. Leverage complementary human–AI abilities — deploy AI where it excels; preserve human strategic veto and contextual judgment; design for safe recomposition.

V. Build open, correctable systems — surface override handles; design for full decision auditability; version every rule set; auto-flag when correction loops are bypassed.

VI. Explore widely, then curate for clarity — generate broad options before narrowing; reveal ranking criteria; offer "show alternates"; discard gracefully.

Full Core Tenets: /about (rendered) or /about/CORE_TENETS.md (raw markdown)

---

## Framework Engine Demo

The Framework Engine is demonstrated in a video: Gen (an AI) helps a user build a structured learning framework from two Winnie-the-Pooh chapters — Chapter 6 (Eeyore's Birthday) and Chapter 9 (Piglet Surrounded by Water). The user is designing a short professional course on teamwork. Together they extract the mechanics of how the characters solve problems (roles, decision-making, showing up with what you have) and name the theme Collaborative Resourcefulness: making something work with what you have, as a team.

Outputs from the demo: (1) Collaborative Resourcefulness Framework — a structured learning tree with competencies and sub-skills. (2) Collaborative Resourcefulness Criteria — assessment criteria aligned to that framework. (3) Full transcript of the conversation. (4) Source chapter texts.

Resources (all at ezralabsai.dev):
- Transcript: /demos/framework-engine/transcript-framework-demo-00.txt
- Framework (plain text): /demos/framework-engine/Collaborative-Resourcefulness-Framework.txt
- Criteria (plain text): /demos/framework-engine/Collaborative-Resourcefulness-Criteria.txt
- Framework PDF: /demos/framework-engine/Collaborative-Resourcefulness%20Framework.pdf
- Criteria PDF: /demos/framework-engine/Collaborative%20Resourcefulness%20Criteria.pdf
- Chapter 6: /demos/framework-engine/chapter-06-eeyores-birthday.txt
- Chapter 9: /demos/framework-engine/chapter-09-piglet-surrounded-by-water.txt

The demo video is embedded on the main page at /#framework-engine.

---

## Architecture

Director → Framework Engine → Assessment Engine → Planning Engine (in sequence; each feeds the next) + management dashboards → central database → Quilt MCP Server + third-party LLM agent → learners. Authoring (Director + engines) is separate from delivery (MCP + agent). One Director, many learners.

## Key URLs

- Main page: /
- About (Mission + Core Tenets): /about
- Mission: /about/MISSION.md
- Core Tenets: /about/CORE_TENETS.md
- Framework Engine: /#framework-engine
- Assessment Engine: /#assessment-engine
- Planning Engine: /#planning-engine
- Quilt MCP Server: /#quilt-mcp-server
- LLM overview (PDF mirror): /llms.pdf