Skip to content

Repository files navigation

/model-intel - AI Model Launch Intelligence Skill for Claude Code

Day-one intelligence gathering for newly released AI models (LLMs, generative media). Produces comprehensive VIBES + DEETS reports within hours of launch.

Example Output

What This Does

Launch 4-5 parallel research agents to capture real-time reactions to new AI model releases:

  • 🐦 Twitter/X sentiment with engagement metrics
  • 💬 Community reactions (HN, Reddit, forums)
  • 📊 Technical benchmarks (independent validation)
  • 🏢 Enterprise testimonials and production metrics
  • 🎯 Competitive positioning analysis
  • ❓ Key unknowns for future investigation

Time to first report: 3-5 minutes after running the command.

Quick Start

1. Install

Paste both lines into Claude Code:

/plugin marketplace add elrolio/skills
/plugin install model-intel@elrolio

Or drop the skill in by hand:

git clone https://github.com/elrolio/model-intel-skill.git /tmp/mis
cp -R /tmp/mis/skills/model-intel ~/.claude/skills/

2. Run

claude "/model-intel GPT-5 Pro"

3. Get Your Report

Output saved to: intel/competitive/[model-name-slug]-launch-intelligence-[date].md

Example Output

See real intelligence report from Claude Opus 4.6 launch: examples/example-output-opus-4.6.md

Includes:

  • Morning pulse (1-3 hrs post-launch)
  • Afternoon pulse (8-16 hrs post-launch)
  • Complete VIBES + DEETS analysis
  • 50+ cited sources

Installation Tiers

Tier Setup Time Quality Cost/Run Dependencies
Minimal 5 min 60% $0.20-0.50 Built-in tools only
Recommended 15 min 95% $0.50-1.50 + Rube MCP (Twitter API)
Full Stack 30 min 100% $0.50-1.50 + Notion + Memory MCP

Start with Recommended for best results.

Documentation

Requirements

  • Claude Code (any recent version)
  • Internet connection
  • Recommended: Rube MCP for Twitter API access (setup)
  • Optional: Notion MCP for publishing, Memory MCP for context

Usage Examples

# Initial morning pulse (within 8 hrs of launch)
claude "/model-intel Claude Opus 4.6"

# Afternoon update (12-16 hrs later)
claude "/model-intel afternoon update for Claude Opus 4.6. morning report: [URL]"

# Test your setup
claude "/model-intel test"

What You Get

VIBES (Social Sentiment):

  • What people are hyped about (themes by frequency)
  • What people are concerned about (with attribution)
  • Community pulse (HN, Reddit, forums)

DEETS (Technical Analysis):

  • Verified performance strengths (benchmarks)
  • Critical technical limitations
  • Enterprise validation (real production metrics)
  • Breaking changes and migration notes

Plus: Competitive positioning, strategic implications, key unknowns, full sources.

Example Report Structure

# Claude Opus 4.6 Launch Intelligence (Feb 5, 2026)

## Executive Summary
[3 paragraphs: standout capabilities, major concerns, positioning]

## THE VIBES
- What People Are Hyped About
- What People Are Concerned About
- Community Forums

## THE DEETS
- Technical Specifications (table)
- Benchmarks Performance (table)
- Enterprise Validation (table)
- Breaking Changes

## Competitive Positioning
[vs GPT-5, vs Gemini, vs Open Source]

## Key Unknowns
[5-7 questions for future investigation]

## Sources
[Organized by category: Official, News, Technical, Community, Enterprise]

See complete example.

Contributing

Fork this repo and customize for your needs:

  • Modify research sources
  • Adjust output format
  • Add custom agents
  • Change report structure

Pull requests welcome for:

  • New research sources
  • Better dependency instructions
  • Alternative MCP integrations
  • Output format improvements

Cost Expectations

Per intelligence run (with Rube MCP):

  • Research agents: ~$0.50-1.50
  • Token consumption: 100K-200K tokens
  • Afternoon updates: ~50% of initial run cost

Composio free tier: 1,000 API actions/month (plenty for model tracking)

Version History

  • 1.0 (2026-02-06): Initial release
    • 5-agent parallel research protocol
    • VIBES + DEETS output structure
    • Rube MCP integration for Twitter/X
    • Afternoon pulse update support
    • Example output from Opus 4.6 launch

License

MIT License - feel free to fork, modify, and share.

Credits

Developed for strategic intelligence gathering on AI model launches. Methodology optimized for Product/Engineering teams conducting evals and Marketing teams tracking narrative.

Maintained by: @elrolio

Support


Ready to install? → Start with INSTALL.md

About

Day-one intelligence gathering for AI model launches - Claude Code skill with parallel research agents

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors