skills.rest
AlexYedi
AlexYediCOMMUNITY
Explore all skills by this author →
GitHub

marketing-autoresearch

Generates and scores marketing copy variants with a simulated expert panel to iteratively optimize conversion content.

1|Updated Apr 9, 2026
Search Tags:#landing page#copywriting#linkedin#conversion optimization#anthropic#a/b testing#variant scoring
One-click install
npx skills add https://github.com/AlexYedi/Empire_State_Events_Pipeline_Take_3 --skill marketing-autoresearch-alexyedi
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: marketing-autoresearch
Source: https://github.com/AlexYedi/Empire_State_Events_Pipeline_Take_3/tree/main/.claude/skills/marketing-autoresearch
Command: npx skills add https://github.com/AlexYedi/Empire_State_Events_Pipeline_Take_3 --skill marketing-autoresearch-alexyedi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires anthropic.

What problem does it solve? Choosing the best headline, CTA, or email subject line normally requires live A/B tests and weeks of traffic. This Skill replaces that with a simulated 5-expert scoring panel that generates dozens of variants, scores them, and evolves winners in minutes — no traffic or test infrastructure needed. ## Core Features & Use Cases - Iterative Variant Optimization: Generates 10+ variants per content element, scores each with a 5-persona expert panel (CMO, skeptical founder, CRO expert, copywriter, founder voice), and evolves top performers across up to 3 rounds. - Multi-Element Cross-Breeding: Combines winning hooks, insights, and CTAs into complete units and scores them holistically. - Full Experiment Logging: Outputs the optimized content, a JSON experiment log with all scores, and a human-readable optimization report. - Use Case: Optimize a LinkedIn post before publishing — run the autoresearch loop on the hook, insight line, and CTA until the average panel score hits 80+, then ship the winning version. ## Quick Start Run autoresearch on my landing page draft to generate and score variants until the copy reaches a score of 85.

Frequently Asked Questions about marketing-autoresearch

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

FAQPage Schema
How do I optimize landing page copy without A/B testing?▼

Run the autoresearch script with your landing page file as input. It generates 10 variants per element, scores them with a simulated 5-expert panel, and evolves winners across rounds until the target score is reached, outputting the best version plus a full experiment log.

How to generate and score multiple headline variants with Claude?▼

The autoresearch.py script batches variant generation and scoring into single Anthropic API calls. Each round produces 10 variants scored 0-100 by five simulated personas, with the average determining the winner.

What content types does autoresearch support?▼

It supports landing pages, email sequences, ad copy, form pages, and LinkedIn posts. Each type has predefined optimizable elements and scoring dimensions, and the type can be auto-detected or forced with the --type flag.

Does simulated expert scoring replace real A/B testing?▼

No. Simulated scoring is for pre-launch optimization only. The skill explicitly recommends running autoresearch first to reach a score of 80+, then deploying and validating with real engagement data.

Why does autoresearch stop improving after a few rounds?▼

If scores stay below 80 after 3 rounds, the issue is strategic rather than tactical — the angle or insight is weak, not the wording. The skill caps runs at 3 rounds by default and recommends revisiting the source brief instead of adding more rounds.

Related Skills

Explore More Marketing & Sales Skills→
mlamla123mlamla123

copywriting

Generates marketing copy for web pages and emails from product briefs.

Community
Intermediate
Mattyreed1Mattyreed1

content-publish

Automate LinkedIn publishing and collect 7-day performance analytics into Notion.

Community
Advanced
propane-aipropane-ai
4

brand-voice

Review content for brand voice, style guide, and messaging compliance.

Official
Intermediate
AlphaBitCoreAlphaBitCore
21

test-cursor-adapter

Validate Cursor adapter protobuf normalization through a synthetic GetChatRequest proxy flow.

Official
Advanced