Materia
A local-first application that turns source material into structured, traceable courses with optional generated audio.
- Period
- Aug 2026 — present
- Use
- Local learning
- Repository
- Public · Apache-2.0
- Technologies
- Next.js · TypeScript · MCP · Python · Docker · pnpm
Courses connected to their sources.
Materia keeps courses and standalone lessons in a local library. Content can start from pasted material or be researched and authored by a compatible agent through the bundled skill and Materia MCP. Each lesson combines semantic chapters, references, comprehension checks, and saved progress.
- 01
Author
Direct lessons from source text or traceable courses authored through an agent and MCP.
- 02
Study
Semantic chapters, cited references, comprehension checks, and local progress.
- 03
Listen
Browser speech, OpenAI TTS, or compatible local voice nodes, always as optional paths.
Explanatory reconstruction
From sources to study.
Four steps summarise how Materia researches, structures, presents, and narrates a course.
Research official sources
The learning goal comes first; the agent then gathers primary material through compatible MCP sources.
Microsoft Learn
Product concepts and guided documentation.
AWS Documentation
Service behavior and technical references.
Google for Developers
Developer guides and API knowledge.
Result: a set of inspectable sources ready to structure into a course.
Build a traceable course
An included authoring skill guides the agent; Materia MCP imports the draft and exposes it again for review.
- 01 Mental model2 objectives · 3 references
- 02 System in motion3 objectives · 4 references
- 03 Decisions and trade-offs2 checks · evidence linked
- Coverage
- Reviewed
- Evidence
- Traceable
- Oral flow
- Checked
Before publication: every imported lesson is read back to check depth, structure, evidence, and teachability.
Study and keep your place
Courses and standalone lessons combine semantic chapters, source references, comprehension checks, and persisted local progress.
How the parts cooperate
A concrete scenario introduces the idea before the lesson moves into definitions and trade-offs.
Go to cited sourceChoose how to listen
Listening is an optional path. Browser speech, OpenAI, and federated local voice nodes stay distinct and explicit.
Browser
Local speech demo with no provider credentials.
OpenAI
Text and audio generation when explicitly configured.
Local node
A compatible voice node operated under local control.
Generation starts only after an explicit confirmation. Jobs are durable and idempotent across refreshes or interruptions.
Awaiting confirmationMateria, in use.
Four captures show the local library, the study view, agent-assisted authoring, and optional provider setup.
01 / 04
Courses and standalone lessons remain available in a local library.
Local-first library and deterministic demo mode