Daily Info Digest

An AI-powered content curation system that tracks builders in AI and delivers a daily digest of their insights.

Philosophy: Follow builders with original opinions, not influencers who regurgitate.

Project Context

Before making changes, read these files:

• ROADMAP.md — current phase and planned features
• CONTEXT.md — shared language and domain terminology
• CLADUE.md — coding behavior guidelines
• docs/adr/ — architectural decisions

File Structure

Daily-Info-Digest/
├── ROADMAP.md              # Three-phase roadmap
├── CONTEXT.md              # Project glossary and domain language
├── CLAUDE.md               # AI coding guidelines
├── SKILL.md                # This file
├── docs/
│   └── adr/                  # Architecture Decision Records
├── web/
│   ├── app.py              # Flask dashboard + report generator
│   ├── templates/          # Jinja2 templates
│   ├── static/             # CSS, JS, generated reports
│   └── data/
│       └── sources.json    # User-managed source list
├── scripts/
│   ├── generate-feed.js   # Fetches X/Twitter, blogs, podcasts
│   ├── prepare-digest.js  # Prepares data for consumption
│   └── deliver.js         # Telegram/email delivery
├── prompts/               # LLM prompt templates
└── config/                # JSON schemas and defaults

Daily Digest Generation Flow

1. generate-feed.js → Fetches raw content from all sources
2. prepare-digest.js → Aggregates and structures the data
3. app.py (build_report_html) → Generates bilingual HTML report
4. Static report served via Flask dashboard

Commands

Generate today's digest:

cd /root/workspace/agent-workspace/Daily-Info-Digest/web
python3 app.py generate

Restart web service:

sudo systemctl restart follow-builders-web

Test report locally:

curl -s http://127.0.0.1:25520/latest

Key Technical Details

• X API: Uses X API v2 with Bearer Token. Fetches tweets, media attachments, and quoted tweets.
• RSS: Parses RSS feeds with regex (no XML parser dependency). Filters by AI keywords.
• Translation: Google Translate free endpoint (translate.googleapis.com). Falls back to English on failure.
• Reports: Self-contained HTML files in web/static/reports/YYYY-MM-DD.html
• Dashboard: Flask on 127.0.0.1:25520, nginx reverse proxy to fb.x-nuwa.com:8445

Adding a New Source

1. Determine source type: rss, hackernews, x, blog, or podcast
2. For RSS: add to web/data/sources.json under media_sites
3. For X: add to web/data/sources.json under x_accounts
4. For blog/podcast: add to web/data/sources.json under appropriate key
5. Test fetch: python3 app.py generate
6. Verify item appears in timeline

Environment

• VPS with root privileges (be cautious)
• Port 443 is reserved for Xray Reality VPN — never touch it
• GitHub repo: Mine77/agent-workspace (project under Daily-Info-Digest/)