AI360Trading — Complete System Documentation

Last updated: March 2026 Read this file before making ANY changes to this project.

1. Project Overview

AI360Trading is a fully automated content + trading signal system running on GitHub Actions (free, public repo = unlimited minutes).

What it does every day — automatically:

• Posts stock buy/sell signals to Telegram (trading bot)
• Creates and uploads 2 YouTube full videos (market analysis + education)
• Creates and uploads 3 YouTube Shorts (ZENO reel + trade setup + market pulse)
• Writes and publishes 4 articles to website
• Shares all content links to Facebook Page + Group
• Sends Instagram caption file for manual posting

Total daily content: 9 pieces — fully automatic Only manual task: Instagram Reels posting (~30 seconds/day)

2. Complete File Map

Root files

| File | Purpose | Called by | |—|—|—| | trading_bot.py | Stock signal scanner — Nifty200 stocks | main.yml | | generate_analysis.py | Part 1: 12-slide market analysis video (~15 min) | daily-videos.yml | | generate_education.py | Part 2: Education video (~12 min) | daily-videos.yml | | generate_shorts.py | Short 2 (trade setup) + Short 3 (market pulse) | daily-shorts.yml | | generate_articles.py | 4 daily market articles pushed to _posts/ | daily-articles.yml | | generate_reel.py | ZENO 60s Hinglish moral reel with Ken Burns effect | daily_reel.yml | | upload_youtube.py | Uploads ZENO reel to YouTube Shorts | daily_reel.yml | | upload_facebook.py | Uploads ZENO reel video to Facebook Page | daily_reel.yml | | upload_instagram.py | Saves Instagram caption to artifact for manual posting | daily_reel.yml | | content_calendar.py | Weekly education topic rotation (Mon–Fri topics) | generate_education.py | | requirements.txt | Python dependencies for all scripts | all workflows | | SYSTEM.md | This file — complete system documentation | humans + AI |

Do NOT delete

• content_calendar.py — generate_education.py imports it directly
• requirements.txt — all workflows depend on it
• public/image/ — ZENO character images used by generate_reel.py and generate_shorts.py
• SYSTEM.md — project memory for AI assistants and developers

3. Workflows (.github/workflows/)

main.yml — AlgoTradeBot

• Runs: Every 5 min, 8:15 AM–4:30 PM IST, Mon–Fri
• Also: 8:45 AM backup, 12:28 PM backup, 3:15 PM backup
• Does: Runs trading_bot.py — scans Google Sheets, sends Telegram signals
• Posts to: Telegram (3 channels: free, advance, premium)
• Timeout: 4 minutes per run

daily-videos.yml — Two YouTube Videos

• Weekday cron: 30 0 * * 1-5 = 6:00 AM IST Mon–Fri
• Weekend cron: 30 3 * * 6,0 = 9:00 AM IST Sat–Sun
• Sequence:
  1. generate_analysis.py → 12-slide market video → uploads to YouTube → saves output/analysis_video_id.txt
  2. generate_education.py → education video → uploads to YouTube → links back to Part 1 → saves output/education_video_id.txt
  3. Facebook share step → posts both video links to Page + Group
• Weekend mode: No live market data — evergreen educational content
• Timeout: 90 minutes

daily-shorts.yml — Two YouTube Shorts (Short 2 + Short 3)

• Weekday cron: 0 6 * * 1-5 = 11:30 AM IST Mon–Fri
• Weekend cron: 0 8 * * 6,0 = 1:30 PM IST Sat–Sun
• Runs after: daily-videos.yml so output/analysis_video_id.txt exists
• Sequence:
  1. Fetch live market data (Nifty, BTC, Gold, S&P500, USD/INR)
  2. Short 2 — trade setup vertical card + Hinglish voice → upload YouTube Shorts
  3. Short 3 — market pulse vertical card + Hinglish voice → upload YouTube Shorts
  4. Facebook share → both short links posted to Page + Group
• Timeout: 30 minutes

daily-articles.yml — 4 Articles

• Weekday cron: 45 1 * * 1-5 = 7:15 AM IST Mon–Fri
• Weekend cron: 30 4 * * 6,0 = 10:00 AM IST Sat–Sun
• Sequence:
  1. generate_articles.py → pushes articles to _posts/ folder
  2. Facebook share step → posts article links to Page + Group
• Weekend mode: Educational/beginner topics instead of market analysis

daily_reel.yml — ZENO Reel (Short 1)

• Weekday cron: 0 3 * * 1-5 = 8:30 AM IST Mon–Fri
• Weekend cron: 30 5 * * 6,0 = 11:00 AM IST Sat–Sun
• Sequence:
  1. generate_reel.py → Groq script → edge-tts voice → Ken Burns video → saves output/reel_YYYYMMDD.mp4
  2. upload_youtube.py → uploads to YouTube Shorts
  3. upload_facebook.py → uploads reel to Facebook Page
  4. upload_instagram.py → saves caption to output/instagram_caption.txt
  5. Artifact upload → reel_*.mp4 + instagram_caption.txt downloadable from Actions UI
• Weekend mode: Emotional/life lesson topics, no market data
• Timeout: 45 minutes

keepalive.yml — Repository Keepalive

• Prevents GitHub from disabling scheduled workflows on inactive repos

4. Content Schedule

Monday to Friday (market days)

| Time IST | Workflow | Content | Posts to | |—|—|—|—| | 6:00 AM | daily-videos | Part 1: Market analysis video ~15 min | YouTube | | 6:20 AM | daily-videos | Part 2: Education video ~12 min | YouTube | | 6:35 AM | daily-videos | Both video links shared | Facebook Page + Group | | 7:15 AM | daily-articles | 4 market articles + Facebook share | Website + Facebook | | 8:30 AM | daily_reel | Short 1: ZENO 60s moral reel | YouTube Shorts + Facebook + Instagram caption | | 11:30 AM | daily-shorts | Short 2: Trade setup clip + Short 3: Market pulse | YouTube Shorts + Facebook |

Saturday and Sunday (market closed)

| Time IST | Workflow | Content | |—|—|—| | 9:00 AM | daily-videos | Educational videos — no live market data | | 10:00 AM | daily-articles | Educational/beginner articles | | 11:00 AM | daily_reel | Short 1: Emotional/life lesson ZENO reel | | 1:30 PM | daily-shorts | Short 2 + Short 3: Educational market content |

Daily content totals

• 2 full YouTube videos (~27 min total)
• 3 YouTube Shorts (~2.5 min total)
• 4 website articles
• 5 Facebook posts (2 video links + 1 article + 1 reel + 1 shorts)
• 1 Instagram caption (manual posting)
• Stock signals every 5 min on Telegram (market hours)

Education topic rotation (content_calendar.py)

| Day | Topic category | |—|—| | Monday | Options and Derivatives | | Tuesday | Technical Analysis | | Wednesday | Global Macro + Fundamentals | | Thursday | Trading Strategies | | Friday | Psychology + Risk Management | | Weekend | Rotates from Monday topics |

5. GitHub Secrets

Important: META_ACCESS_TOKEN expires every 60 days

Refresh process:

1. Go to developers.facebook.com/tools/explorer
2. Select AI360Trading app → Generate Access Token
3. Open: https://graph.facebook.com/me/accounts?access_token=NEW_TOKEN
4. Copy the access_token from response
5. Update META_ACCESS_TOKEN secret in GitHub
6. Set phone reminder every 50 days

6. ZENO Character System

Image files (in public/image/)

| Filename | Emotion | Used when | |—|—|—| | zeno_happy.png | Happy | Positive topics, bullish market | | zeno_celebrating.png | Excited/celebrating | Strong positive, CTA ending | | zeno_thinking.png | Curious/thinking | Educational, reflective topics | | zeno_sad.png | Sad | Difficult topics, bearish market | | zeno_fear.png | Worried/fear | Cautionary lessons, market fear | | zeno_angry.png | Angry | Strong warning topics | | zeno_greed.png | Greedy | Money/greed related topics |

Story arc logic in generate_reel.py (3 images per reel)

Every reel uses 3 ZENO images — never random, always tells a story:

• Image 1 (hook): matches script emotion
• Image 2 (middle/lesson): related reflective emotion
• Image 3 (CTA/end): always positive — zeno_celebrating or zeno_happy

Example for sad topic: zeno_sad → zeno_thinking → zeno_celebrating

ZENO in generate_shorts.py

• Short 2: uses zeno_happy or zeno_celebrating (bullish) / zeno_sad or zeno_fear (bearish)
• Short 3: same mood-based selection from public/image/

Ken Burns animation (generate_reel.py only)

Each image gets smooth zoom or pan:

• Segment 1: zoom_in
• Segment 2: pan_left
• Segment 3: zoom_out

7. Pipeline Diagrams

ZENO Reel (Short 1) Pipeline

GitHub Actions trigger (8:30 AM IST)
→ Weekend detection (datetime.weekday() >= 5)
→ Groq API → Hinglish script (weekday: market+moral, weekend: emotional)
→ edge-tts → Hindi voice (hi-IN-SwaraNeural)
→ select_zeno_images() → 3 images based on emotion + market mood
→ make_ken_burns() → zoom/pan effect on each image
→ moviepy → background + ZENO segments + voice + subtitles + title
→ output/reel_YYYYMMDD.mp4
→ upload_youtube.py → YouTube Shorts
→ upload_facebook.py → Facebook Page Reel
→ upload_instagram.py → saves instagram_caption.txt

Shorts 2 + 3 Pipeline

GitHub Actions trigger (11:30 AM IST)
→ Read output/analysis_video_id.txt (link to today's Part 1)
→ fetch_market() → Nifty + BTC + Gold + S&P500 + USD/INR (Yahoo Finance)
→ Short 2:
   make_short2_frame() → vertical 1080x1920 trade setup card (PIL)
   make_short2_script() → Groq Hinglish trade setup script
   edge-tts → hi-IN-SwaraNeural voice
   moviepy → short2_YYYYMMDD.mp4
   YouTube API → upload as Short
→ Short 3:
   make_short3_frame() → vertical 1080x1920 market pulse card (PIL)
   make_short3_script() → Groq Hinglish market pulse script
   edge-tts → hi-IN-MadhurNeural voice (different voice = variety)
   moviepy → short3_YYYYMMDD.mp4
   YouTube API → upload as Short
→ Facebook → share both short links to Page + Group

Analysis + Education Videos Pipeline

GitHub Actions trigger (6:00 AM IST)
→ generate_analysis.py:
   fetch_all() → Nifty + BankNifty + 9 global + Options + FII/DII
   Groq → 12 analysis slides JSON
   PIL → render each slide 1920x1080 PNG
   edge-tts → voice for each slide
   moviepy → analysis_video.mp4 (~15 min)
   YouTube API → upload → save analysis_video_id.txt
   PIL → custom thumbnail
→ generate_education.py:
   content_calendar.py → get today's topic
   read analysis_video_id.txt → link Part 1
   Groq → education slides JSON
   PIL → render slides
   edge-tts → voice
   moviepy → education_video.mp4 (~12 min)
   YouTube API → upload → update Part 1 with Part 2 link
→ Facebook → share both video links

8. Known Fixes Applied

9. Platform Details

YouTube

• Short 1 (ZENO reel) → Shorts (vertical 1080x1920, ~60s)
• Short 2 (trade setup) → Shorts (vertical 1080x1920, ~45s)
• Short 3 (market pulse) → Shorts (vertical 1080x1920, ~35s)
• Analysis video → regular video (horizontal 1920x1080, ~15 min)
• Education video → regular video (horizontal 1920x1080, ~12 min)
• Channel: @ai360trading
• Auth: YOUTUBE_CREDENTIALS JSON secret (OAuth2 refresh token — never expires)

Facebook

• Page: “Unofficial Amit Kumar” (will rename to ai360trading after followers)
• Group: ai360trading
• ZENO reel → uploaded as Facebook Reel via video_reels API
• Articles + video + shorts links → feed posts with link preview
• Token: META_ACCESS_TOKEN — expires every 60 days, must refresh manually

Instagram

• Cannot auto-post Reels via API (Meta restriction as of 2026)
• Caption saved to output/instagram_caption.txt in GitHub Actions artifact
• Download artifact → copy caption → post manually (~30 seconds)

Telegram

• 3 channels: free (ai360trading), advance (Rs499/month), premium (bundle)
• ONLY for stock buy/sell signals — not used for video/reel/article notifications
• AlgoTradeBot runs every 5 min during market hours

Website

• URL: https://ai360trading.in
• Static Jekyll site on GitHub Pages
• Articles auto-pushed to _posts/ by generate_articles.py
• RSS: https://ai360trading.in/feed/

10. Monetization Status and Roadmap

Current status

• YouTube Partner Program: working toward 1,000 subs + 4,000 watch hours
• AdSense: not yet applied
• Facebook monetization: working toward 10,000 followers
• Paid Telegram: Rs499/month advance channel active

Earning targets

| Milestone | Target month | Expected earning | |—|—|—| | AdSense approved | Month 4-6 | Rs3,000-15,000/month | | 50 paid Telegram subs | Month 3-4 | Rs25,000/month | | 100 paid Telegram subs | Month 5-6 | Rs50,000/month | | Facebook monetization | Month 6-8 | Rs5,000-20,000/month | | Live trading (Angel One) | After 30 paper trades | Variable |

11. Future Roadmap (Phase 3+)

• Twitter/X auto-post — daily Nifty + BTC + reel link
• Reddit auto-post — share articles to trading communities
• YouTube Community Posts — auto-post daily poll + Nifty levels
• Pinterest auto-pin — article links for US/Brazil/UK traffic
• Live trading via Angel One SmartAPI (after 30 paper trades validated)
• Instagram auto-posting (blocked by Meta API — check again in future)
• ZENO lip-sync animation (requires GPU — not possible on free GitHub Actions)
• Facebook Page rename to ai360trading (after more followers)
• META_ACCESS_TOKEN auto-refresh automation

12. Quick Reference — All Cron Times

13. How to Test Each Workflow

1. Go to GitHub repo → Actions tab
2. Click workflow name on left sidebar
3. Click “Run workflow” → branch: master → Run workflow
4. Watch steps live — green = success, red = error
5. Click any red step → expand logs → find error message

Recommended test order (first time or after changes)

1. daily_reel.yml — simplest, tests ZENO + YouTube + Facebook
2. daily-shorts.yml — tests Short 2 + Short 3
3. daily-articles.yml — tests article generation + Facebook share
4. daily-videos.yml — longest (~30 min), test last
5. main.yml (AlgoTradeBot) — already running, do not break

14. How to Use This File with Any AI Assistant

Paste this prompt to any AI:

“Read the SYSTEM.md file in my GitHub repo ai360trading. It explains the complete automation system. I want to [your request here]. Make sure you don’t break anything that’s already working.”

The AI will understand the full system instantly. Always update this file after major changes.

Complete AI360Trading automation system — built with Claude AI, March 2026. 9 pieces of content daily. Zero cost. Fully automatic. The journey from zero to earning — documented here.