autoresearch

Generate and score marketing content variants using a multi-persona rubric.

3.3k|656|Updated Mar 28, 2026
One-click install
npx skills add https://github.com/ericosiu/ai-marketing-skills --skill autoresearch-ericosiu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: autoresearch
Source: https://github.com/ericosiu/ai-marketing-skills/tree/main/autoresearch
Command: npx skills add https://github.com/ericosiu/ai-marketing-skills --skill autoresearch-ericosiu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires anthropic.

What problem does it solve?

Autoresearch removes guesswork in marketing copy by generating multiple versions, scoring them against a simulated expert panel, and evolving the winners across rounds until you reach a strong conversion-oriented draft.

Core Features & Use Cases

  • Karpathy-style optimization loop: Generate 10+ variants, score with a 5-persona panel, then evolve top performers through multiple rounds.
  • Conversion-focused scoring: Produces a structured score breakdown using dimension-specific rubrics for landing pages, emails, ad copy, and form pages.
  • Cross-breeding for complete units: Combines winners across elements to create holistic versions and selects the best overall option.
  • Experiment logging: Outputs the optimized content plus a full experiments JSON log and a human-readable optimization report.

Quick Start

Run the command to optimize your landing page: python3 autoresearch.py --input landing-page.html --type landing_page --min-score 85

Frequently Asked Questions about autoresearch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I optimize landing page copy without real traffic experiments?

To optimize landing page copy without real traffic, you can generate multiple variants, score them with a simulated multi-persona panel, and iteratively evolve the highest performers. This removes guesswork by relying on structured rubrics instead of live A/B testing.

What is AI-driven conversion copy scoring and how does it work?

AI-driven conversion copy scoring evaluates marketing content by generating multiple variants and rating them against a 5-persona expert panel using dimension-specific rubrics. Top performers are cross-bred across multiple rounds to produce a highly optimized final draft.

How do I generate and test email sequences before launching a campaign?

You can generate and test email sequences by running an iterative optimization loop that creates batched variants, scores them with a multi-persona rubric, and evolves the winners. This allows messaging refinement during pre-launch preparation without needing live traffic.

Do I need an Anthropic API key to run AI copy optimization loops?

Yes, you need a valid ANTHROPIC_API_KEY to run these AI copy optimization loops. The Anthropic API dependency is required for batched variant generation and single-call scoring during each optimization round.

Can I use AI to refine ad copy and form pages without A/B testing?

Yes, you can refine ad copy and form pages by replacing live A/B testing with a Karpathy-style optimization loop. It generates variants, scores them against a structured rubric, and outputs an optimized artifact along with experiment logs.

What's the best way to iterate on marketing messaging using an AI rubric?

The best way to iterate on marketing messaging is to generate 10+ variants, score them with a 5-persona panel, and cross-breed winners across multiple rounds. This produces a holistic, optimized artifact plus a human-readable summary report.