Lunaris

Turn a topic into a real, verified course.

Name a topic. Lunaris plans it, writes the lessons, and verifies every claim against evidence. You watch each step run.

AGPL-3.0·Every claim cited·Taught in order·Self-hostable
lunaris · How binary search works RELEVANCE
  1. Scoping relevance Modeling what you already know; bounding the build to the gap. RELEVANCE
  2. Planning the curriculum Decomposing into 5 knowledge components. PLAN
  3. Mapping prerequisites Topologically ordering concepts · checking the graph is acyclic. GRAPH
  4. Researching sources Running 12 queries · scoring credibility & trust tier. RESEARCH
  5. Writing the lessons Merrill phases, activate to demonstrate to apply to integrate. AUTHOR
  6. Grounding every claim Checking each factual sentence against retrieved evidence. VERIFY
  7. Publishing Cut 3 unsupported claims · course ready to read. PUBLISH
How it works
01 · How it works

From a topic to a course you can trust, in one run.

One canvas, four moves. You name it; the agent does the rest in the open, and you read the result.

Name a topic Type what you want to learn. Lunaris models your level and scopes the build to the gap, nothing you already know.
Watch it plan The agent decomposes the topic, orders prerequisites into an acyclic graph, and shows every step as it runs.
Grounded authoring It researches sources, writes Merrill-structured lessons, and verifies each claim against the evidence it found.
Read & learn A reader with objectives, claims that carry their sources, and a prerequisite map you can explore freely.
02 · Order

Never taught before its prerequisites.

Every topic is decomposed into concepts and assembled into a directed acyclic graph, ordered topologically in deterministic code. The model proposes; the graph builder proves.

  • Cycle detection rejects circular dependencies before authoring.
  • Nodes carry a difficulty tier that colors the Map.
  • The order is proven, not prompted.
lunaris · prerequisite graph ✓ acyclic
Comparison Arrays Loops Sorted order Binary search
tier 1, foundation tier 3 tier 5, goal
03 · Truth

Unsupported claims never ship.

A claim-level verifier checks every factual sentence against retrieved evidence. What survives carries its source, a trust tier, and a credibility score. What doesn't is cut before publish.

23 claims 20 kept 3 cut
lunaris · grounding verifier 20 kept · 3 cut
Binary search runs in O(log n) on a sorted array. official96%
Each step halves the remaining search interval. reputable88%
Binary search always beats a hash lookup. cut · no support
It needs random access to the middle element. official91%
04 · The course

What comes out is a real artifact.

Merrill-structured lessons, objectives, and claims that carry their sources. This is the Reader, on the course this page has been building.

lunaris · How binary search works · lesson 5 of 5 published
Lesson 05 · binary search

Halving a sorted range

objective: apply binary search to locate a target in O(log n)

“A sorted range lets you discard half the remaining candidates with a single comparison.”

Cormen et al., Introduction to Algorithms §2.3 official94%

Probe the middle. If the value is too small, the answer lives in the right half; too large, the left. Either way, half the candidates are gone.

23 claims verified every claim carries its source
05 · Get started

What do you want to learn?

Open the studio and name a topic, or run the whole stack yourself with one command.

macOS · Linux · Windows
# fresh clone -> installs everything, starts the stack, opens the studio
$ make run

No keys? Lunaris falls back to a deterministic build and says so.