English | 中文

AI Intelligence System (AIS)

Live demo: contents.voxlinkai.com

An OpenClaw skill package that builds a fully automated AI news intelligence pipeline — collecting, analyzing, notifying, reporting, and querying — all powered by LLM-driven agent skills.

What It Does

17:00  ais-collect   → Fetches from RSS, Hacker News, X/Twitter, GitHub Trending
17:30  ais-analyze   → LLM clusters items into scored events
18:00  ais-notify    → Sends Telegram alerts for critical events (score ≥ 7)
19:00  ais-report    → Generates & emails a daily HTML report
 any   ais-query     → Natural language Q&A over your intelligence database

Skills

Skill	Trigger	Description
ais-collect	Cron 17:00	Multi-source data collection (RSS, HN, X, GitHub)
ais-analyze	Cron 17:30	LLM-powered event clustering, scoring (1-10), categorization
ais-notify	Cron 18:00	Telegram instant alerts for high-score events
ais-report	Cron 19:00	Daily HTML/text report via email (Resend SMTP)
ais-query	On demand	Natural language knowledge base queries

Quick Start

Prerequisites

• OpenClaw installed and configured
• Python 3.8+
• SQLite3
• (Optional) Resend account for email reports
• (Optional) Telegram bot via OpenClaw for notifications

1. Install skills

Clone the repository:

git clone https://github.com/axzml/ai-intelligence.git ~/ai-intelligence

Then install the skills into OpenClaw using one of the two methods below.

Method A: Copy to OpenClaw's default skills directory

cp -r ~/ai-intelligence/skills/ais-* ~/.openclaw/workspace/skills/

Verify the skills are recognized — ask OpenClaw:

What skills do you have? Do you see ais-collect, ais-analyze, ais-notify, ais-report, ais-query?

OpenClaw should list all 5 ais-* skills. If not, check that the directory structure is correct: each skill folder should contain a SKILL.md file.

Method B: Add a custom skills path in openclaw.json

Edit your OpenClaw config file (usually ~/.openclaw/openclaw.json) and add the skills field pointing to this repo's skills directory:

{
  "skills": {
    "load": {
      "extraDirs": ["/home/<YOUR_USER_NAME>/ai-intelligence/skills"],
      "watch": true
    }
  }
}

Replace <YOUR_USER_NAME> with your actual system username. The watch: true option makes OpenClaw auto-reload when skill files change.

2. Initialize the database

python scripts/init_db.py

3. Configure data sources

Edit config/rss_feeds.json to customize RSS feeds, HN keywords, X/Twitter accounts, and GitHub trending filters.

4. Configure settings

Edit config/settings.json:

• scoring.user_interests — your interest tags for relevance scoring
• schedule — adjust pipeline timing
• webapp.domain — your web dashboard domain (if deploying the webapp)

5. Set up email (optional)

sqlite3 ~/ai-intelligence/data/intelligence.db
UPDATE config SET value = 'your-resend-api-key' WHERE key = 'resend_api_key';
UPDATE config SET value = 'noreply@yourdomain.com' WHERE key = 'email_sender';
UPDATE config SET value = 'you@example.com' WHERE key = 'email_recipient';

6. Set up cron jobs

bash scripts/setup_cron.sh

Or manually add to crontab (times in your server timezone):

0 17 * * * openclaw run ais-collect
30 17 * * * openclaw run ais-analyze
0 18 * * * openclaw run ais-notify
0 19 * * * openclaw run ais-report

7. Deploy web dashboard (optional)

See deploy/SERVER_DEPLOY_GUIDE.md for Nginx + systemd setup.

cd webapp
pip install -r requirements.txt
python app.py

Project Structure

├── skills/                  # OpenClaw agent skills (the core)
│   ├── ais-collect/         # Data collection from 4 sources
│   ├── ais-analyze/         # LLM analysis & event generation
│   ├── ais-notify/          # Telegram notifications
│   ├── ais-report/          # Daily email reports
│   └── ais-query/           # Natural language queries
├── config/
│   ├── settings.json        # System configuration
│   └── rss_feeds.json       # Data source definitions
├── scripts/
│   ├── init_db.py           # Database initialization
│   └── setup_cron.sh        # Cron job installer
├── heartbeat/
│   └── HEARTBEAT.md         # Health-check checklist for monitoring
├── webapp/                  # Flask web dashboard
└── deploy/                  # Server deployment configs

Heartbeat (Health Monitoring)

The heartbeat/HEARTBEAT.md defines a checklist that OpenClaw runs periodically to monitor system health:

• Detects unprocessed item backlog (raw_items piling up)
• Alerts if high-score events are pending notification
• Warns if the daily report failed to generate
• Reports data source fetch errors

To enable it, append the heartbeat checklist to OpenClaw's workspace:

cat ~/ai-intelligence/heartbeat/HEARTBEAT.md >> ~/.openclaw/workspace/HEARTBEAT.md

OpenClaw will automatically pick up HEARTBEAT.md and run the checks on its configured heartbeat cycle. You can also trigger it manually by asking OpenClaw: "Run the heartbeat check."

Scoring System

Events are scored 1-10 using weighted criteria:

Weight	Criterion	Description
25%	Source Authority	Official blogs score higher than personal posts
25%	Interest Match	Relevance to your configured interests
20%	Novelty	New concepts score higher than incremental updates
15%	Community Heat	Engagement metrics (upvotes, likes, comments)
15%	Timeliness	Breaking news scores higher than old items

License

MIT