SkillNet turns what you want to teach into courses, grounded tutors, learning materials and adaptive interfaces.
For one learner, a class, a team or an organization.
Learning does not always begin inside an organization or an existing document. Sometimes it starts with a topic you want to understand. Sometimes the knowledge already lives in PDFs, manuals, notes or conversations.
SkillNet supports both paths. Describe what you want to learn or teach, or upload PDF, DOCX, Markdown and TXT material. It then builds a structured course with lessons, exercises, a tutor and learning media. Uploaded material remains the grounding source. Courses that begin from an idea preserve the provenance of their generated source instead of presenting it as uploaded evidence.
The knowledge, objectives and evaluation criteria stay stable. The explanation, example, activity, medium and interface can change with the learner's preferences, experience, current state and bounded interaction signals.
| The learning contract | The experience can adapt |
|---|---|
| Knowledge and sources | Explanations and examples |
| Objectives and evidence | Practice and support |
| Evaluation criteria | Medium, sequence and interface |
Responding to a request in the moment is not the same as learning what has helped a person over time. SkillNet treats those signals as revisable evidence, not fixed “learning styles” and not a claim that the system already knows the learner perfectly.
- Creates complete courses from a topic or PDF, DOCX, Markdown and TXT sources.
- Answers with a course tutor that retrieves enrolled sources and returns provenance.
- Composes learning experiences with OpenUI and a supported subset of Didact with a pinned version.
- Generates learning media such as podcasts, infographics, slide decks and narrated videos when the corresponding providers are configured.
- Records progress and skills through enrollments, attempts, mastery and explicit verification.
- Works at different scales through individual and organization workspaces, from personal study to classes, teams and larger deployments.
- Connects to other tools through its REST API and optional A2A and MCP adapters.
- Runs on your infrastructure with Docker and an OpenAI-compatible model provider.
Try SkillNet in the live demo, run it locally or read the documentation.
cp .env.example .env
docker compose up -d --build
docker compose exec api python -m src.seed_learning_demo # optional public demoOpen http://localhost:3000. The full running guide covers provider configuration, keyless fixtures, demo data and troubleshooting.
- Vision: why learning software should adapt to people.
- Product: current scope and product direction.
- Roadmap: the next four priorities.
- ANFAIA release snapshot: what this version contains.
- OpenUI adoption: the controlled GenUI runtime.
- Didact integration: how learning components enter SkillNet.
- Contributing: development setup, checks and conventions.
SkillNet is the main project. Didact, OpenUI, mcp-md-reader, A2TL-Web, A2TL-Video, Curio and DBP explore related parts of the same direction at different levels; they are not all dependencies of the current runtime.
SkillNet is open source under the Apache 2.0 license. Security issues should follow SECURITY.md, never a public issue.