Support the Docs¶
Every contribution helps
If this documentation saved you time — or helped you get your radio on the air faster — your support keeps the pipeline running and the assistant improving.
Built for a serious application¶
AetherSDR is not a weekend project. It is built with C++20 and Qt6 from day one — the best technology available for an application that demands real-time DSP performance, low latency, and native cross-platform behavior on Linux, macOS, and Windows. That level of engineering deserves documentation that matches it.
I believe documentation is a first-class component of any software project — not an afterthought. Good docs reduce friction for new users, help experienced operators get the most out of every feature, and lower the barrier to entry for a hobby that already has a steep enough learning curve.
An AI-powered documentation pipeline¶
This manual is not written by hand page by page. It is generated by a custom pipeline that I built to keep pace with AetherSDR releases:
- Source ingestion — with each new release, the pipeline pulls the AetherSDR C++ source code directly from the repository.
- AI analysis — a large language model scans every user-facing object exposed by the application: windows, controls, menus, settings, and behaviors.
- High-quality output — the model produces structured, accurate documentation in natural English, covering what each feature does, how to use it, and relevant configuration options.
The result is a living manual that evolves with the software — not a static document that goes stale two releases later.
Athena — your AI assistant¶
Every page on this site includes Athena, an AI assistant you can open at any time by clicking the chat button in the bottom-right corner.
Athena is powered by RAG — Retrieval-Augmented Generation. Here is what that means in plain language:
Instead of relying solely on what a language model learned during training, RAG first retrieves the most relevant passages from a real knowledge base — in this case, the actual documentation — and then generates an answer grounded in that content.
This means Athena does not hallucinate features that do not exist or describe behavior from a different software version. Her answers are anchored to the current AetherSDR documentation, with source links so you can verify everything she says.
Under the hood:
- Vector database — documentation chunks are embedded and stored for fast semantic search
- Voyage AI embeddings — state-of-the-art text embeddings for retrieval accuracy
- Claude (Anthropic) — the language model that synthesizes retrieved context into clear answers
- FastAPI backend — deployed on Railway for reliable, low-latency responses
- Advanced mode — toggle ⚙️ to also search AetherSDR source code, not just the docs
Athena is continuously improving. New documentation, better retrieval tuning, and smarter prompts ship with every iteration.
Open source, wherever possible¶
This project leans on open source at every layer:
| Component | Technology |
|---|---|
| Documentation site | MkDocs Material |
| Markdown rendering in chat | marked.js + DOMPurify |
| Vector store | ChromaDB |
| API framework | FastAPI |
| Hosting | GitHub Pages (docs) + Railway (backend) |
Infrastructure has real costs¶
GitHub Pages hosts the documentation for free. Everything else does not:
- Vector database — fine-tuned embeddings and persistent storage for every AetherSDR release
- LLM API calls — every question Athena answers costs tokens
- Voyage AI embeddings — high-quality semantic search requires a paid embeddings API
- Railway deployment — always-on backend for the assistant
If you find this documentation useful — if it saved you an hour of digging through forums, or helped you get your radio on the air faster — your support genuinely helps keep this system running and makes it better over time.
Documentation pipeline and AI assistant built by LU5DX. AetherSDR is developed by ten9876.