
My Focal AI · AI 资讯研究台
编辑介绍
按个人关注主题形成可持续更新的 AI 新闻桌面。
AI 用途:AI 聚合、筛选并总结作者关注的新闻与研究线索。
使用方式:本地模式接用户 Ollama;云端评分可使用 Anthropic API;也可用 --no-score 完全跳过 AI。
原帖正文
MyFocalAI
Personal news intelligence system that aggregates AI content from curated sources (RSS, YouTube, Twitter, arXiv, events, GitHub trending, AI tools trending, agent skills), scores relevance using LLM, and serves a web dashboard. Runs locally with Ollama (free) or deployed to Vercel with Claude API scoring.
| Feeds | Admin |
|-------|-------|
|
|
|
Quick Start
git clone https://github.com/YanCheng-go/my-focal-ai.git
cd my-focal-ai
./start.sh # installs deps, starts RSSHub + Ollama, launches dashboard
Open http://localhost:8000 — you're done.
./start.sh --no-score # skip Ollama (just fetch + display)
./start.sh stop # stop all services
Prerequisites: uv and Docker. Optionally Ollama for LLM scoring.
Deployment Modes
| Mode | Stack | Description | |------|-------|-------------| | Local | SQLite + Ollama + FastAPI | Full-featured, runs on your machine | | Online Public | Vercel + GitHub Actions | Static read-only dashboard, no backend needed | | Online Login | Supabase + Vercel | User accounts, per-user feeds, on-demand fetch |
See docs/deployment.md for setup instructions.
Documentation
- docs/deployment.md — deployment modes, setup, configuration
- docs/development.md — dev setup, project structure, commands, conventions
- docs/architecture.md — data flow, design decisions
- docs/module-map.md — file-by-file module map
- docs/sources.md — how to configure and add sources
- docs/scoring.md — scoring principles, tiers, LLM prompt
Support
If you find this project useful:
Last updated: 2026-03-22
来源与作者
- 原作者:Yan Cheng / 程彦 (@YanCheng-go)
- 发布平台:GitHub
- 原帖:查看原帖
- 权利说明:VibeMySpace 仅作带来源的整理展示,内容与图片权利归原作者。
开源信息
未开源。