RISC-V Radar is a small static intelligence briefing for public RISC-V signals: news, papers, companies, markets, and engineering activity. It is designed to run daily on GitHub Actions and publish with GitHub Pages.
export PYTHONPYCACHEPREFIX=/tmp/riscv-radar-pycache
python scripts/collect.py
MINIMAX_API_KEY=your_key python scripts/translate.py
python scripts/build_site.py
python -m http.server 8080 --directory distOpen http://localhost:8080.
- RSS and Atom feeds from RISC-V, semiconductor, Linux, and embedded sources.
- arXiv metadata through the public arXiv API.
- GitHub repository metadata through GitHub Search.
- Market sources are configured as optional placeholders. Use a reliable finance API such as Alpha Vantage, Finnhub, or Polygon before enabling automated market collection.
The collector stores normalized records in data/daily/YYYY-MM-DD.json and the static site in dist/.
The translation step adds Chinese titles, summaries, and editorial takeaways while preserving the original source fields and links.
The Obsidian export creates obsidian-vault/daily/YYYY-MM-DD.md and updates obsidian-vault/index.md for local indexing and search.
For GitHub Actions, configure the repository secret MINIMAX_API_KEY. The workflow uses the China-region endpoint https://api.minimaxi.com/v1 and model MiniMax-M3.
Open /Users/xiebinbb/github/riscv-radar/obsidian-vault/ as an Obsidian vault. After each daily workflow, pull the repository and Obsidian will index the new Markdown note locally.
To build the newest local note manually:
python scripts/build_obsidian.pyEdit config/sources.json. Prefer RSS or official APIs before direct website scraping.
- Create an empty GitHub repository, for example
riscv-radar. - Push this project to the repository.
- In repository settings, enable GitHub Pages with GitHub Actions as the source.
- Run the
Daily RISC-V Radarworkflow manually once.
The workflow is scheduled at 20 0 * * * UTC, roughly 08:20 in Beijing time.
Keep the site as a source-linked briefing, not a reposting site. Store titles, URLs, short public summaries, metadata, and your own commentary. Avoid copying full articles.