Multisource Intel Radar

Build and run a high-signal information radar for C-end founders and operators across YouTube, X/Twitter, Reddit, WeChat Official Accounts, and Xiaohongshu....

Create a founder-grade signal system: less junk, more decisions.

Inputs

• OPML file (default: /Users/rogeryang/Downloads/follow.opml)
• Keyword whitelist (default: 创业, AI, 增长, 金融)
• Optional source lists for non-RSS channels (X list links, subreddit list, WeChat/XHS accounts)

Output Contract

Always output:

1. Top 3 must-read signals (one-line why + one action + clickable source link)
2. Top 5 watchlist items (with source link)
3. Dropped noise summary (what got filtered and why)
4. Filter transparency (counts + rates: scanned -> matched -> shortlisted -> top3)
5. Next experiment (one concrete growth/ops move)

Workflow

Step 1) Ingest feed sources

• Parse OPML with scripts/parse_opml.py
• Generate normalized feed list: assets/feeds.txt

Step 2) Fetch + filter

• Run scripts/build_digest.py with keyword whitelist
• Time window default: last 48h
• Keep only items that match whitelist in title/summary
• For Xiaohongshu: do browser search (not watchlist-dependent), using keyword combos like:
  • 创业 AI 增长
  • AI 产品 复盘
  • 增长运营 案例 Then append top findings with profile/note links.

Step 3) Score items

Use this weighted scoring:

• Relevance to whitelist (40%)
• Actionability in 7 days (30%)
• Novelty / non-obviousness (20%)
• Evidence density (10%)

Step 4) Summarize for execution

For each selected item, provide:

• Core insight (1 sentence)
• Why it matters for current product
• Suggested action today (1 step)

Source Coverage Notes

• YouTube/X/Reddit often available via RSSHub or platform feeds in OPML
• WeChat OA and Xiaohongshu are often not natively RSS; add via:
  • RSS bridge links (if available)
  • Manual watchlist files (assets/wechat_watchlist.txt, assets/xhs_watchlist.txt)
• If a source has no feed, include it in watchlist and mark as manual scan required

Anti-Noise Rules

• Do not output generic motivational posts
• Drop repeated观点 without new evidence
• Prefer first-hand data / concrete case over opinion
• Keep digest under 10 items total

Daily Cadence (recommended)

• 09:30: morning digest (strategic)
• 18:30: evening digest (tactical)

Keyword Defaults

创业, AI, 增长, 金融

If user provides new keywords, merge and deduplicate.

Files

• Parser: scripts/parse_opml.py
• Digest builder: scripts/build_digest.py
• Notes: references/scoring-and-ops.md