Proven by data, not vibes

Make any AI skill
measurably better

Optimize prompts, skills, and instructions for Claude, ChatGPT, Cursor, Gemini — any AI. Get a report card proving the improvement.

Optimize a skill — $25Free starter pack
+16%
avg. pass rate improvement
-20%
avg. prompt size reduction
5/5
binary eval criteria
4
output formats

Same prompt. Different results.

We tested the “brainstorming” skill 10 times each. Here's what changed.

Original — 80% pass rate

What went wrong

  • Asked 4 questions at once instead of 1
  • Generic "what's your purpose?" opener
  • Proposed approaches with no trade-offs
  • Never mentioned Python, slides, or speaker notes
  • Could have been a response to any project
Augmented — 96% pass rate

What improved

  • One focused question per message
  • Referenced the user's Python preference immediately
  • 3 approaches with clear trade-offs for each
  • Anchored to syntax highlighting and speaker notes
  • Felt built for THIS request, not a template

How it works

1

Submit your skill

Paste a link to any .md skill file, or upload it directly. Works with Claude Code skills, CLAUDE.md files, system prompts — anything.

2

We optimize it

50 mutation cycles with binary evaluation. We test, mutate, keep what's better, discard what's worse. Then benchmark 10x to prove it.

3

You get the proof

Optimized skill + MCP server + system prompt + report card. Works in Claude, ChatGPT, Cursor, Windsurf, VS Code, Gemini.

Pre-optimized skills

Ready to use. Report card included. Works in every AI tool.

FREE
🧠

Brainstorming

Turns ideas into designs through structured dialogue. One question at a time. Multiple approaches with trade-offs.

80% → 96%Free
FREE
📋

Writing Plans

Creates bite-sized TDD implementation plans with exact file paths, runnable commands, and real code snippets.

80% → 96%Free
🔍

Code Review

Catches bugs, security issues, and logic errors with severity levels. Acknowledges what's done well, not just problems.

+22% avg$25
🐛

Bug Fixing

Systematic debugging — logs first, root cause analysis, verify fix. Never guesses. Never band-aids.

+18% avg$25

Test Writing

Generates meaningful tests, not boilerplate. Edge cases, error paths, integration scenarios. TDD ordering enforced.

+24% avg$25
📝

SEO Content

Write content that ranks. Keyword integration, proper headings, internal linking suggestions, meta descriptions.

+28% avg$25
💰

Sales Copy

Landing pages, email sequences, ad copy. Benefit-driven, objection-handling, with clear CTAs.

+20% avg$25
📖

API Documentation

Generates clear, complete API docs with examples, error codes, and auth flows. Developers actually read these.

+16% avg$15
🔄

Refactoring

Clean code transformations that preserve behavior. Extracts patterns, reduces duplication, improves naming.

+15% avg$15
Download free starter pack

Includes brainstorming + writing plans with originals for side-by-side comparison

Prefer to run it yourself?

$5 one-time. Your skills never leave your machine.

# Install nothing — runs via npx
npx skill-augment init

# Choose your mode
npx skill-augment run --cli # free with Claude Code sub
npx skill-augment run # ~$10 via API key

# Bundle all optimized skills
npx skill-augment bundle
Get the DIY tool — $5

Works with Claude Code, accepts GitHub URLs, generates MCP servers