Draft specificationCore + extensions

A portable snapshot of what a person knows.

LSL is an open standard for learner state. It gives AI systems, tutors, and learning platforms a shared way to represent what a learner currently understands, where the gaps still are, and how confident those conclusions should be.

Read the intro Read the specification

Diagram showing LSL as a learner state snapshot rather than event logs

Document preview

# The smallest valid LSL document.
# Core layer only. Empty lists are valid — thin-but-honest beats rich-but-fabricated.

lsl: "1.0"
learner: "lrn-3f9qm"
updated: 2026-07-12

concepts:
  - name: "CSS Flexbox"
    mastery: 0.55
    confidence: low
    last_seen: 2026-07-10
    can:
      - "Centers items in a single-row container"
    cannot_yet: []
    misconceptions: []

{
  "lsl": "1.0",
  "learner": "lrn-3f9qm",
  "updated": "2026-07-12",
  "concepts": [
    {
      "name": "CSS Flexbox",
      "mastery": 0.55,
      "confidence": "low",
      "last_seen": "2026-07-10",
      "can": [
        "Centers items in a single-row container"
      ],
      "cannot_yet": [],
      "misconceptions": []
    }
  ]
}

LEARNER: lrn-3f9qm · updated 2026-07-12
MASTERY: CSS Flexbox 0.55 low (stale 2d)
CAN: Centers items in a single-row container

What to know

A compact state file, built for AI and humans alike.

It describes what a learner knows right now, and nothing more.

Core

State, not events

Current mastery, confidence, freshness, and next steps.

Reading model

Plain-language claims

Readable by an LLM without a lookup table.

Extensions

Layered when needed

Provenance, relationships, and federation stay optional.

Next step

Open the spec or inspect the examples.

Open the specification Browse examples