Reverse-design learning. Tell Telos what you want to master, and it builds the dependency graph, diagnoses where you are, teaches only the missing edges, and schedules review.
Live demo · Open the app · 中文 README
Most learning tools start from content. Telos starts from the outcome.
- Goal to graph: break an outcome into prerequisite knowledge points.
- Diagnosis first: find the learner's actual starting point before teaching.
- Just-enough lessons: explain, quiz, repair, and verify each missing node.
- Review loop: keep mastered nodes alive with FSRS-style scheduling.
- Local-first Community Edition: run the product app and reference runtime on your own machine.
Use the hosted Telos at telos.ungetsu.net for cloud sync, managed workflows, and the official deployment.
git clone https://github.com/YunyueLi/telos.git
cd telos
./start.shThen open http://localhost:3000. The web app automatically connects to the local runtime at 127.0.0.1:8787.
For the AI parts, copy core/.env.example to core/.env and add an OpenAI-compatible key:
TELOS_LLM_API_KEY=your-api-key
TELOS_LLM_BASE_URL=https://api.deepseek.com
TELOS_LLM_MODEL=deepseek-v4-proRun the engine tests:
make testRun with Docker:
docker compose up --build- Personal mastery maps for any technical or academic goal.
- Course generators that adapt to a learner's current knowledge state.
- Study agents that use Telos Core as the planning and review engine.
- Self-hosted learning products with local-first data and BYOK model access.
- Research prototypes around knowledge tracing, prerequisite graphs, and review scheduling.
| Area | What it contains |
|---|---|
landing/ |
Static marketing page served at /. |
web/ |
Next.js app served at /app/, with map, diagnosis, review, studio, settings, and local-first project state. |
core/ |
Zero-dependency Python learning engine and local reference runtime. |
skill/ |
Agent skill package for using Telos inside external agent workflows. |
scripts/ |
Static export and local build helpers. |
docs/ |
Product design notes, roadmap, and deployment guide. |
The Community Edition is intentionally runnable: the product UI, local graph workflow, reference server, examples, tests, and docs live here in the canonical public repository.
Issues and pull requests are open for the Community Edition. Good contributions usually improve:
- learning-engine correctness and tests
- graph/data contracts
- local runtime reliability
- web app usability
- examples, docs, and self-hosting paths
Read CONTRIBUTING.md, DEPLOYMENT.md, and BRAND.md before opening larger changes.
Telos Community Edition is licensed under AGPL-3.0. The standalone core/ package is additionally provided under Apache-2.0 for SDK/protocol-level reuse. See NOTICE for boundaries.