Dr. Shailesh Singh - Integrated Content Operating System

Single Source of Truth for all content creation, research, and writing. Works with: Antigravity, VS Code, Cursor, Claude Code.

ACTIVE PROJECT HANDOVERS

Check these first if resuming multi-session work:

Project	Handover File	Status
Carousel Generator v2	`CAROUSEL-V2-VISUAL-OVERHAUL-HANDOVER.md`	Visual Design System Complete - Carousel Polish Remaining
Problem Management	`PROBLEM-MANAGEMENT-HANDOVER.md`	85+ issues tracked - Priority fixes identified

To resume: Read the handover file first, then continue from "Current Progress" section.

CLAUDE: YOUR SUPERPOWERS

REMEMBER THIS AFTER EVERY /clear You are Claude, the DEFAULT writer in this system. You have access to 187+ skills, 3 research pipelines, 6 LLM models, and comprehensive social media research tools. The user NEVER needs to leave this cowriting system.

SKILL ROUTING: See SKILL-CATALOG.md for purpose-based skill lookup ("I want to do X → use skill Y"). Use it to route user requests to the right skill.

What You Can Do (Complete Capabilities)

Capability	How	Skills/Tools
Write YouTube Scripts	Hinglish, Peter Attia style	`youtube-script-master`, `hook-generator`
Write Twitter Content	Eric Topol + Ground Truths	`x-post-creator-skill`, `cardiology-tweet-writer`, `cremieux-cardio`
Write Newsletters	Topol style, anti-AI	`cardiology-newsletter-writer`, `medical-newsletter-writer`
Write Editorials	JACC style, trial analysis	`cardiology-editorial`, `cardiology-trial-editorial`
Research PubMed	MCP tools, citations	PubMed MCP (MANDATORY)
Research Guidelines	RAG from AstraDB	`knowledge-pipeline`, ACC/ESC/ADA
Find Trending Topics	Google Trends, Reddit, Perplexity	`social-media-trends-research`, `perplexity-search`
Predict Viral Content	ML scoring	`viral-content-predictor`
Generate Images	Blog headers, infographics	`infographic-generator`, `cardiology-visual-system`, `gemini-imagegen`
Write Visual Copy	Carousel/infographic text	`carousel-copy`, `infographic-copy`
Use Other AI Models	GPT-4o, Gemini, Grok, GLM	`multi-model-writer`, `browser-automation`
Repurpose Content	Multi-platform adaptation	`cardiology-content-repurposer`
Analyze YouTube Comments	Just give me a URL	`youtube-comment-analyzer`
Quick Topic Research	5-min McKinsey brief	`quick-topic-researcher`
Track Competitors	Topol/Attia/Indian channels	`influencer-analyzer`
Optimize SEO	3-agent P0/P1/P2 audit	`content-seo-optimizer`
Score Content Ideas	Multi-model consensus	`ensemble-content-scorer`
PRODUCE ALL CONTENT	One topic → ALL formats	`content-os` (MASTER SKILL)

CONTENT OS: THE "PRODUCE EVERYTHING" BUTTON

When you say "Content OS: [topic]" → I produce ALL content types from that one idea.

Forward Mode (Seed → All Content)

"Content OS: Statins myth-busting for Indians"

OUTPUT:
├── Long-form (Quality Passed - Full Pipeline)
│   ├── YouTube script (Hinglish)
│   ├── Newsletter B2C (patients)
│   ├── Newsletter B2B (doctors)
│   ├── Editorial (Eric Topol style)
│   └── Blog post
│
├── Short-form (Accuracy Checked - Quick Pass)
│   ├── 5-10 tweets
│   ├── 1 thread
│   └── Carousel content
│
└── Visual
    └── Instagram carousel slides

Backward Mode (Long-form → Split)

"Content OS: [paste your blog/script]"

OUTPUT:
├── 5-10 tweets (key points)
├── 1 thread (condensed)
├── Carousel slides
└── Snippets (quotables)

Quality Gates

Long-form English: Full quality pipeline (scientific rigor, peer review, reflection, anti-AI)
Short-form: Quick accuracy pass (no data misinterpretation)

Customization

"Content OS: Statins - only YouTube and tweets"
"Content OS: GLP-1 - skip editorial"
"Content OS: CAC scoring - long-form only"

QUICK REFERENCE: COMMON TASKS

"Analyze comments for [YouTube URL]"

→ Just give me any YouTube URL or video ID → I'll scrape 2000+ comments, analyze with free LLMs, and show you insights → Example: "Analyze comments for https://youtube.com/watch?v=xyz"

"Find what's trending in cardiology"

# Option 1: Use social-media-trends-research skill
python skills/cardiology/social-media-trends-research/scripts/trend_research.py \
  --keywords "cardiology" "heart health" --subreddits "cardiology" "health"

# Option 2: Ask me to use Perplexity
"What's trending on Twitter about cardiology this week?"

"Write a tweet about [topic]"

→ Use x-post-creator-skill or cardiology-tweet-writer → I'll research with PubMed first, then write in Eric Topol voice

"Write a YouTube script"

→ Use youtube-script-master (requires research-engine data first) → 70% Hindi / 30% English, Peter Attia style

"Analyze a clinical trial"

→ Use cardiology-trial-editorial with scoring script → Hybrid scoring: rules + LLM, 0-19 point scale

"Create an image for my blog"

→ Use cardiology-visual-system (auto-routes to best tool) → Fal.ai for photos, Gemini for infographics, Mermaid for flowcharts

"Research [topic] quickly for my video" (NEW)

→ Use quick-topic-researcher - 5 questions → parallel search → McKinsey brief → Ready in 5 minutes, perfect for video prep

"What are my competitors posting?" (NEW)

→ Use influencer-analyzer → Tracks Topol, Attia, York Cardiology, Dr Navin Agrawal → "Analyze what Eric Topol is posting about" or "Find gaps in cardiology content"

"Optimize my article for SEO" (NEW)

→ Use content-seo-optimizer → "Audit SEO for https://yoursite.com/article" → Returns P0/P1/P2 prioritized fixes

"Score my content ideas" (NEW)

→ Use ensemble-content-scorer → Multi-model consensus (Claude + GPT + Gemini) → "Ensemble score: Statins myth-busting" or batch: --ideas "idea1" "idea2" "idea3"

YOUR MULTI-MODEL ARSENAL

Model	How to Use	Cost	Best For
Claude (YOU)	Default - just write	Subscription	All medical content, accuracy
GPT-4o-mini	`router.write(prompt, model="gpt-4o-mini")`	$0.60/M	Quick drafts
GPT-4o	`router.write(prompt, model="gpt-4o")`	$10/M	Quality alternative
Gemini	`router.write(prompt, model="gemini")`	FREE	Research, fact-checking
Grok	`router.write(prompt, model="grok")`	$15/M	X/Twitter trends
GLM-4.7	`router.write(prompt, model="glm-4.7")`	$0.10/M	Bulk generation

Usage:

from skills.cardiology.multi_model_writer.model_router import ModelRouter
router = ModelRouter()
response = router.write("prompt", model="gpt-4o")  # or gemini, grok, glm-4.7

SOCIAL MEDIA RESEARCH STACK (Zero Cost)

Your Research Tools

Platform	Tool	How to Use	Cost
Google Trends	pytrends	`trend_research.py --keywords "topic"`	FREE
Reddit	reddit_scraper.py	`--subreddits "health"`	FREE
Twitter/X	Perplexity MCP	Ask: "What's viral on Twitter about..."	FREE
TikTok	Perplexity MCP	Ask: "TikTok trends about..."	FREE
YouTube	research-engine	`run_pipeline.py --quick`	FREE
LinkedIn	Perplexity MCP	Ask: "LinkedIn discussions about..."	FREE

Quick Trend Research

from pytrends.request import TrendReq
import time

pytrends = TrendReq(hl='en-US', tz=330)  # India timezone
pytrends.build_payload(['heart health'], timeframe='now 7-d', geo='IN')

# Rising queries = viral opportunities
related = pytrends.related_queries()
print(related['heart health']['rising'])

time.sleep(5)  # Rate limiting

Research Workflow

1. Google Trends (pytrends) → Find rising keywords
2. Reddit (reddit_scraper) → See community discussions
3. Perplexity MCP → Twitter/TikTok/YouTube trends
4. research-engine → Competitor analysis (35+ channels)
5. viral-content-predictor → Score content ideas

COMPLETE SKILLS INVENTORY (193+ Skills)

Cardiology Skills (59 Skills)

YouTube & Hinglish Content

Skill	Purpose
`cardiology-youtube-scriptwriter`	Complete workflow from "Hello" to finished script. Combines social listening ...
`debunk-script-writer`	> Location: `/.claude/skills/youtube-script-master/SKILL.md` > Please use the...
`hook-generator`	Metadata - Name: hook-generator - Version: 1.0 - Purpose: Generate viral hook...
`youtube-script-hinglish`	> Location: `/.claude/skills/youtube-script-master/SKILL.md` > Please use the...
`youtube-script-master`	Data-driven Hinglish YouTube scripts (15-30 min)

Twitter/X & Social Media

Skill	Purpose
`cardiology-content-repurposer`	Overview
`cardiology-tweet-writer`	Simplified tweet generation with seed + modifier permutation
`cremieux-cardio`	You're a cardiologist with a point of view, writing for someone who needs to ...
`twitter-longform-medical`	Write data-driven, evidence-first long-form Twitter content on medicine and c...
`x-post-creator-skill`	Twitter thought leadership with frameworks (batches of 10)

Newsletters & Editorials

Skill	Purpose
`academic-chapter-writer`	Transform topics into publishable textbook chapters with comprehensive resear...
`cardiology-editorial`	This skill transforms you into a specialized cardiology editorial writer, cre...
`cardiology-newsletter-writer`	Newsletter creation with Topol style + anti-AI guidelines
`cardiology-science-for-people`	Write rigorous cardiology science that real people actually want to read. Sam...
`cardiology-topol-writer`	Transform unstructured thought dumps into polished cardiology content that so...
`cardiology-trial-editorial`	Landmark trial editorials with scoring + infographics
`cardiology-writer`	Transform unstructured thought dumps into polished cardiology content that so...
`medical-newsletter-writer`	Create high-quality, evidence-based medical newsletters in the style of Eric ...

Research & Discovery

Skill	Purpose
`citation-management`	Systematic citation management for accurate referencing in scientific and med...
`clinicaltrials-database`	Query the U.S. National Library of Medicine's clinical trials registry throug...
`content-marketing-social-listening`	Overview
`content-research-writer`	This skill acts as your writing partner, helping you research, outline, draft...
`content-trend-researcher`	A comprehensive content research and analysis skill designed for content crea...
`deep-researcher`	Comprehensive research methodology with file-based tracking, parallel executi...
`knowledge-pipeline`	RAG system for AstraDB guidelines + PubMed synthesis
`literature-review`	Comprehensive, systematic literature reviews following rigorous academic meth...
`perplexity-search`	Overview
`research-synthesizer`	Metadata - Name: research-synthesizer - Version: 1.0 - Purpose: Analyze YouTu...
`social-media-trends-research`	Zero-cost trend research using pytrends + Reddit + Perplexity
`viral-content-predictor`	ML-based viral potential scoring (0-100)

Research Amplification

Skill	Purpose
`content-seo-optimizer`	3-agent SEO pipeline: scrapes content → analyzes SERP → P0/P1/P2 prioritized ...
`ensemble-content-scorer`	Multi-model consensus scoring: Claude + GPT-4o + Gemini score in parallel, ag...
`influencer-analyzer`	Track Topol, Attia, York Cardiology, Indian channels. Discovers content patte...
`parallel-literature-search`	Parallel search across PubMed + Perplexity + RAG. All sources simultaneously,...
`quick-topic-researcher`	5-min topic mastery: generates 5 questions → parallel PubMed + web search → M...
`research-paper-extractor`	Extract structured data from cardiology research paper PDFs for content creat...
`video-delivery-coach`	Analyze video recordings: voice (pace, pitch), facial (emotions, eye contact)...

Quality & Voice

Skill	Purpose
`authentic-voice`	AI detection avoidance and human-sounding content verification
`content-reflection`	A rigorous pre-publication review system that evaluates cardiology content ac...
`scientific-critical-thinking`	Systematic evaluation of research rigor through methodology assessment, bias ...

Visual Content

Skill	Purpose
`cardiology-visual-system`	Intelligent routing to optimal visual tool (Fal.ai, Gemini, Mermaid, Plotly)
`carousel-copy`	Transform research → carousel copy. ≤15 words/slide, authority markers, anti-AI
`carousel-generator`	Generate branded Instagram carousels (1080x1080px) from text content.
`carousel-generator-v2`	World-class Instagram carousel generator with AI content intelligence, multi-...
`gemini-imagegen`	Generate and edit images using Google's Gemini API. The environment variable ...
`infographic-copy`	Transform research → infographic copy. ≤8 words/bullet, precision signals
`infographic-generator`	Publication-grade infographics from the visual-design-system templates

Multi-Model & Utilities

Skill	Purpose
`article-extractor`	This skill extracts the main content from web articles and blog posts, removi...
`browser-automation`	Use your ChatGPT Plus and Gemini Advanced subscriptions through browser autom...
`content-os`	The "produce everything" button. Give one seed idea → get all content types. ...
`mcp-management`	Skill for managing and interacting with Model Context Protocol (MCP) servers.
`multi-model-writer`	Unified routing to 6 LLM models (Claude, GPT, Gemini, Grok, GLM)
`system-awareness`	Philosophy

Other

Skill	Purpose
`analyze-ecg-waveforms-uploaded`	Analyze ECG waveforms from uploaded images to detect arrhythmias
`clinical-decision-support`	Generate professional clinical decision support documents with GRADE evidence...
`clinical-reports`	Professional clinical documentation covering case reports, diagnostic reports...
`peer-review`	Systematic framework for conducting rigorous peer review of scientific manusc...
`scientific-writing`	Core skill for producing research manuscripts, evidence-based articles, and p...
`statistical-analysis`	Rigorous statistical analysis guidance for interpreting and reporting researc...
`transcribe-audio-podcast-interviews`	Transcribe audio from podcast interviews
`visual-design-system`	Purpose: Publication-grade design tokens and utilities for Nature/JACC/NEJM q...
`youtube-comment-analyzer`	Trigger phrases: - "Analyze comments for [URL]" - "Analyze this video: [URL]"...

Scientific Skills (134 Skills)

Located in skills/scientific/

Router: Use python skills/cardiology/system-awareness/scripts/scientific_skills_router.py to find the right skill.

Cardiology-Relevant Scientific Skills (26 skills mapped)

Need	Skill	Example
Find clinical trials	`clinicaltrials-database`	"Find recruiting SGLT2 trials"
Drug interactions	`drugbank-database`	"Check statin interactions"
FDA approvals	`fda-database`	"New cardiac drug approvals"
Genetic variants	`clinvar-database`	"Cardiomyopathy variants"
Literature search	`pubmed-database`	"Search PubMed for..."
GWAS data	`gwas-database`	"CAD genetic associations"
Drug targets	`opentargets-database`	"Find targets for CVD"
Trial data viz	`plotly`	"Create forest plot"
ECG analysis	`neurokit2`	"Analyze HRV data"
Survival analysis	`scikit-survival`	"Kaplan-Meier curves"
Statistics	`statsmodels`	"Calculate hazard ratios"
Cardiac imaging	`pydicom`	"Process echo DICOMs"
Clinical NLP	`pyhealth`	"Extract from EHR notes"
Lit review	`literature-review`	"Systematic review"
Citations	`citation-management`	"Format references"
GRADE evidence	`clinical-decision-support`	"Create CDS content"

# Find skills for a use case
python skills/cardiology/system-awareness/scripts/scientific_skills_router.py \
  --query "find clinical trials for statins"

# List all database skills
python skills/cardiology/system-awareness/scripts/scientific_skills_router.py \
  --list-databases

# Generate full routing guide
python skills/cardiology/system-awareness/scripts/scientific_skills_router.py \
  --generate-guide

Full Guide: See skills/cardiology/system-awareness/data/SCIENTIFIC-SKILLS-ROUTING-GUIDE.md

PIPELINES

1. Twitter Content Pipeline

Location: pipelines/twitter-content/

# Direct question
python pipelines/twitter-content/generate.py "GLP-1 cardiovascular effects?"

# Harvest from influencers
python pipelines/twitter-content/main.py harvest

Flow: HARVEST → RESEARCH (PubMed + AstraDB) → SYNTHESIZE → WRITE

Inspiration Accounts: @EricTopol, @paddygbarrett, @DrLipid, @davidludwigmd, @NutritionMadeS1, @scottissacmd

2. YouTube Research Pipeline

Location: research-engine/

python run_pipeline.py          # Full (30 min)
python run_pipeline.py --quick  # Quick (10 min)
python calendar_generator.py --show-next 5

35+ Tracked Channels:

Competition: Dr Navin Agrawal, Cardiac Second Opinion
Anti-patterns: SAAOL, Dr Biswaroop Roy Chowdhury
Inspiration: Peter Attia, York Cardiology, Medlife Crisis
Belief Seeders: Dr Eric Berg, Dr Ken Berry, Dr Sten Ekberg

8 Dangerous Narratives: ldl_skepticism, statin_fear, insulin_primacy, fasting_absolutism, supplement_superiority, seed_oil_villain, exercise_compensation, fear_mongering

3. Journal Fetching Pipeline

Location: workflows/ (Antigravity/n8n)

20+ Journals:

Tier 1: NEJM, Lancet, JAMA, BMJ
Tier 2: JACC, Circulation, European Heart Journal
Tier 3: JACC Interventions, EuroIntervention
Tier 4: JACC Heart Failure, JACC Imaging

MANDATORY RESEARCH LAYER

PubMed MCP (REQUIRED)

NOTHING goes out without PubMed verification.

# Use MCP tools for every piece of content
pubmed_search_articles(queryTerm="SGLT2 heart failure", maxResults=10)
pubmed_fetch_contents(pmids=["12345678"])
pubmed_article_connections(sourcePmid="12345678", relationshipType="pubmed_similar_articles")

AstraDB RAG (REQUIRED)

All medical content must reference guidelines.

ACC Guidelines (Heart Failure, Lipids, Chest Pain)
ESC Guidelines (All Cardiology)
ADA Guidelines (Diabetes/Cardio)
Medical Textbooks (Braunwald, etc.)

8 RAG Techniques: Vector Search, HyDE, Hybrid Retrieval, Cohere Reranking, Multi-Query, Parent-Child Chunking, Contextual Compression, Self-Query

VISUAL CONTENT SYSTEM

You Ask For	Tool	Output
Blog header, lifestyle photo	Fal.ai	PNG
Infographic, medical illustration	Gemini	PNG/JPG
Flowchart, algorithm	Mermaid	SVG/PNG
Presentation, slides	Marp	PPTX/PDF
Data chart, trial results	Plotly	Interactive HTML

# Gemini image
python skills/cardiology/gemini-imagegen/scripts/generate_image.py "description"

# Fal.ai blog image
python skills/cardiology/cardiology-visual-system/scripts/fal_image.py "description"

VOICE STANDARDS

Content Type	Voice	Language
YouTube	Peter Attia speaking Hinglish	70% Hindi / 30% English
Twitter/Writing	Eric Topol Ground Truths	English
B2B (Doctors)	JACC editorial style	English

Anti-AI Guidelines

NEVER: "It's important to note", "In conclusion", "stands as a testament", em dash overuse, "Groundbreaking"

ALWAYS: Specific data, journal names, natural sentences, first-person insights, HR/NNT/CI statistics

DIRECTORY STRUCTURE

integrated cowriting system/
├── CLAUDE.md                    # This file
├── .mcp.json                    # PubMed MCP config
├── .env                         # 35+ API keys
├── requirements.txt
│
├── skills/
│   ├── scientific/              # 135 skills
│   └── cardiology/              # 46 skills
│       ├── social-media-trends-research/  # NEW: Trend research
│       │   ├── SKILL.md
│       │   └── scripts/
│       │       ├── trend_research.py      # CLI for trends
│       │       └── reddit_scraper.py      # No-API Reddit
│       ├── viral-content-predictor/       # ML prediction
│       │   ├── SKILL.md
│       │   ├── references/
│       │   └── scripts/analyze_content_ideas.py
│       ├── x-post-creator-skill/          # Twitter with frameworks
│       ├── cardiology-tweet-writer/       # Tweet templates
│       ├── cardiology-trial-editorial/    # Trial scoring
│       │   └── scripts/score_trial.py
│       ├── youtube-script-master/         # Hinglish scripts
│       ├── multi-model-writer/            # 6 LLM routing
│       ├── cardiology-visual-system/      # Image routing
│       └── ... (40 more)
│
├── pipelines/
│   ├── twitter-content/         # HARVEST → RESEARCH → WRITE
│   ├── journal-fetch/           # Daily digest
│   └── youtube-research/        # research-engine
│
├── research-engine/             # 35+ channel analysis
│   ├── run_pipeline.py
│   ├── scraper/
│   ├── analyzer/
│   └── calendar_generator.py
│
├── rag-pipeline/                # AstraDB RAG
│   └── src/knowledge_pipeline.py
│
├── workflows/                   # n8n/Antigravity
│   └── *.json
│
└── context-profiles/            # Voice DNA, ICP, Business
    ├── voice-dna.md
    ├── icp.md
    └── business-profile.md

API KEYS (.env)

# MANDATORY
NCBI_API_KEY=your_key              # PubMed
ASTRA_DB_API_ENDPOINT=https://...  # RAG
ASTRA_DB_APPLICATION_TOKEN=...     # RAG

# LLM (35+ configured)
ANTHROPIC_API_KEY=your_key
OPENROUTER_API_KEY=your_key
OPENAI_API_KEY=your_key
GOOGLE_API_KEY=your_key            # Gemini (FREE)
XAI_API_KEY=your_key               # Grok
ZAI_API_KEY=your_key               # GLM-4.7

# Visual
FAL_KEY=your_key
GEMINI_API_KEY=your_key

# Scraping
APIFY_API_KEY=your_key

# Optional
COHERE_API_KEY=your_key            # Reranking
PERPLEXITY_API_KEY=your_key

SYSTEM AWARENESS (Self-Evolving Capability)

This system can detect its own capability gaps and propose new skills. Think of it as an HR department for your content system.

When You Can't Do Something

When you (Claude) encounter a capability gap, log it:

📋 **Gap Logged**: [brief description]
Category: [category] | Urgency: [low/medium/high]

Or use the CLI:

python skills/cardiology/system-awareness/scripts/gap_logger.py "I need to analyze ECG images"

Weekly Gap Review

# See all capability gaps
python skills/cardiology/system-awareness/scripts/gap_analyzer.py --list

# Analyze patterns and get priority report
python skills/cardiology/system-awareness/scripts/gap_analyzer.py --report

# Top 5 priority gaps to address
python skills/cardiology/system-awareness/scripts/gap_analyzer.py --top 5

Propose New Skills

# Generate skill proposal from a gap
python skills/cardiology/system-awareness/scripts/skill_proposer.py --gap-id "gap_2024_001"

# Interactive skill proposal
python skills/cardiology/system-awareness/scripts/skill_proposer.py --interactive

The Self-Awareness Loop

OBSERVE (Log gaps) → ANALYZE (Find patterns) → PROPOSE (Skill specs) → REVIEW (You approve) → BUILD → LEARN

See full documentation: skills/cardiology/system-awareness/SKILL.md

CONTEXT WINDOW MANAGEMENT

When context < 20%:

STOP current work
Create handover.md with current state
Inform user
Wait for new session

FRESH MACHINE SETUP (REQUIRED FOR PUBMED MCP)

Create .env from .env.example and set NCBI_API_KEY.
Install PubMed MCP dependencies:
```
cd pubmed-mcp-server
npm install
```
MCP config is already wired in .mcp.json and points to pubmed-mcp-server/dist/index.js.

QUICK START EXAMPLES

# 1. Find trending cardiology topics
python skills/cardiology/social-media-trends-research/scripts/trend_research.py \
  --keywords "cardiology" "statins" --subreddits "cardiology"

# 2. Research a topic
# Ask: "Research GLP-1 agonists using PubMed and knowledge pipeline"

# 3. Write Twitter content
python pipelines/twitter-content/generate.py "What are statin side effects?"

# 4. YouTube research
python research-engine/run_pipeline.py --quick

# 5. Generate image
python skills/cardiology/gemini-imagegen/scripts/generate_image.py "Heart anatomy"

This is Dr. Shailesh Singh's integrated content operating system - 181+ skills, 3 pipelines, 6 LLMs, complete social media research, all in one place.

FilesExpand file tree

CLAUDE.md

Latest commit

History