A common language for learning
OLM is a structural framework that makes educational programs portable, inspectable, and reproducible. It provides a governed vocabulary — canonical routines, artifacts, evidence types, patterns, and constraints — so any program can be expressed in a common structure, whether the person doing that work is a human educator or an AI system.
OLM is not a curriculum. It is a layer that sits over curricula and makes their structure visible.
openlearningmap.org · Contribute · Community
program_id: program.homemade_pizza
title: "Homemade Pizza"
embedding_mode: standalone
typical_duration: "60 min"
age_range: "all ages"
patterns:
- pattern.opening
- pattern.middle
- pattern.closing
routines:
- routine.discuss
- routine.new_material
- routine.hands_on_activity
artifacts:
- id: artifact.physical_product
durability: perishable
capture_note: "Pizza is consumed; photo optional."
evidence:
primary:
- evidence.artifact_presence
supporting:
- evidence.routine_completion
constraints:
- constraint.time_limit
- constraint.material_limit
hdd_alignment:
- hdd.curiosity
- hdd.sensemakingEvery ID in every packet is validated against the canonical registry. No generated packet may reference an ID not in canonical/canonical_registry.yaml.
| Category | Count |
|---|---|
| Patterns | 12 |
| Routines | 21 |
| Artifacts | 19 |
| Evidence types | 5 |
| Constraints | 9 |
| HDDs | 8 |
Browse: canonical/canonical_registry.yaml
olm/
├── canonical/ # Canonical element registry + individual YAML files
│ ├── canonical_registry.yaml
│ ├── patterns/
│ ├── routines/
│ ├── artifacts/
│ ├── evidence/
│ ├── constraints/
│ └── hdds/
├── examples/ # Four reference implementations
│ ├── popcorn_factory/ # Authored — venture & making, week-long camp
│ ├── homemade_pizza/ # Authored — culinary, 60-min workshop
│ ├── cad_missions/ # Integrated — technical/CAD, self-paced series
│ └── openscied_6_1/ # Integrated — science inquiry, 6-week unit
├── programs/ # Community-contributed programs (staging → canonical)
├── rituals/ # OLM Ritual Library
├── templates/ # Fillable Core Mapping and Playbook templates
├── docs/ # Framework documents
│ ├── constitution.md
│ ├── hdd_framework.md
│ ├── playbook_guide.md
│ ├── authoring_guide.md
│ └── glossary.md
└── CONTRIBUTING.md
Every OLM program packet has six layers, each serving a distinct audience:
| Layer | Purpose | Audience |
|---|---|---|
| Core Mapping | Learning architecture — programs, patterns, routines, artifacts, evidence | Hub buyers, designers |
| Playbook | Delivery model — pedagogy, educator stance, adaptation rules | Program leads |
| Runbook | Executable session flow — timed steps, facilitator actions | Internal pipeline |
| Educator Brief | What educators see — prep, materials, session flow, key prompts | Educators |
| Parent Brief | What families see — accessible description, what kids will do | Parents |
Core Mappings and Playbooks are the reusable structural artifacts stored in this repo. Runbooks, Educator Briefs, and Parent Briefs are generated output.
Four programs mapped end-to-end in examples/:
Popcorn Factory — The primary reference. Meta Humans–authored venture and making program. Shows all six learning patterns across a week-long camp.
Homemade Pizza — OLM at its simplest. A 60-minute culinary workshop with one perishable artifact. Shows that OLM works at any scale.
CAD Missions — OLM as an interpretive overlay on an external program (cadmissions.com). One Core Mapping covers the full catalog. evidence.numeric_match against a surface-area check is the strongest evidence signal in the library.
OpenSciEd Unit 6.1 — OLM as a translation layer over a widely-used NSF-funded curriculum (openscied.org). OpenSciEd's five "routines" become OLM patterns and a ritual — naming the difference is what makes the translation possible.
OLM is designed to be legible to both humans and AI systems. Its governed vocabulary and explicit layer rules eliminate the most common failure modes in AI-generated educational content: activities masquerading as programs, skills asserted without derivation, evidence claims without artifacts.
When a model operates within OLM, it cannot hallucinate structural relationships. The constraints become guardrails.
Downloadable AI briefs — system prompt, context bundle, and canonical registry — are available at openlearningmap.org/ai.
The most valuable contribution is a Core Mapping — a structural description of a program you actually run, expressed in canonical OLM vocabulary. No GitHub account required: submit through the web form at openlearningmap.org/contribute.
See CONTRIBUTING.md for the full process, or browse templates/ for fillable starting points.
OLM framework documents and canonical registry: LICENSE-CONTENT.md
Built and maintained by Meta Humans.