rachittshah

rachittshah / optimize-anything

Public

Universal text artifact optimizer using LLM-powered iterative search

10
0
100% credibility
Found Mar 07, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A tool that iteratively refines text artifacts such as prompts, code snippets, or configurations by generating AI-proposed improvements and evaluating them against user-defined scoring criteria.

How It Works

1
🔍 Discover the optimizer

You hear about a smart tool that automatically makes your prompts, instructions, or writing better by trying improvements and picking the winners.

2
📝 Pick your starting point

Choose a piece of text to improve, like a chat prompt or set of instructions you want clearer and more effective.

3
⭐ Tell it what good looks like

Describe simply how to judge improvements, such as checking if it's clear, helpful, or works well on examples.

4
▶️ Start the improvement process

Give the starting text and your judgment rules, then launch the tool to begin refining it.

5
⏳ Follow the progress

Watch as it generates better versions, tests them against your rules, and keeps the top performers.

🎉 Grab your perfected version

Celebrate having the best-improved text that's scored highest and ready to use perfectly.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is optimize-anything?

optimize-anything is a Python library and CLI that acts as a github universal text optimizer, iteratively refining any text artifact—prompts, code, configs, or agent instructions—via Claude-powered proposals guided by your evaluator's feedback. Feed it a seed candidate, a scoring function (Python callback, shell command, or LLM judge), an objective like "maximize clarity," and a dataset; it evolves better versions through targeted mutations, tracking a Pareto frontier for multi-task gains. A TypeScript MCP server exposes tools like `optimize_anything` and `get_best_candidate` for Claude Code integration.

Why is it gaining traction?

It stands out by turning vague "optimize anything" goals into structured search with actionable diagnostics fed to Claude, skipping blind evolution for precise fixes. Users get quick wins like boosting a skill from 91% to 100% accuracy, plus flexible evaluators that hook into existing benchmarks without rewriting tests. The event streaming and checkpointing make long runs manageable.

Who should use this?

Prompt engineers tuning LLM instructions across examples, backend devs optimizing configs or scripts via shell evals, AI builders generalizing agent prompts on train/val splits. Suited for Pythonistas with a scorer ready to automate iteration on universal text artifacts.

Verdict

At 10 stars and 1.0% credibility, it's raw but mature in docs, quickstart demos, and full test coverage—try the CLI or API for low-risk experiments. Solid for Claude fans; expect to tweak for production scale.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.