Cranot

Cranot / agi-in-md

Public

System prompts as cognitive lenses. 11 compression levels, 393+ experiments, 19 domains. A 170-word markdown file makes any model derive conservation laws between impossibilities.

76
11
100% credibility
Found Mar 03, 2026 at 76 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

A research project sharing markdown-based system prompts that reliably trigger advanced reasoning patterns in AI models across code, legal, creative, and other domains, supported by hundreds of experiments.

How It Works

1
🔍 Discover the prompts

You stumble upon a collection of simple text guides that unlock smarter thinking in AI chats, perfect for analyzing code, stories, or ideas.

2
đź“– Explore the guide

You read stories of how short notes make AI spot hidden patterns and suggest clever fixes in everyday work like writing or planning.

3
đź’ˇ Pick your thinking tool

You choose a short guide that matches what you need, like naming hidden issues or dreaming up better designs, and copy it easily.

4
đź’¬ Start your AI chat

You open your favorite AI conversation tool and paste in the guide as instructions for smarter responses.

5
đź“ť Share your work

You type or paste the code, poem, contract, or idea you want analyzed, and ask the AI to dig deep.

6
🔥 Watch insights unfold

The AI delivers clear, powerful breakdowns—like revealing trade-offs or impossible fixes—that feel like magic but are reliable.

🎉 Unlock better ideas

Your project improves with fresh perspectives, deeper understanding, and creative solutions you couldn't see before.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 76 to 76 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 agi-in-md?

This repo delivers markdown system prompts as cognitive lenses for LLMs, compressing complex reasoning into 3-235 words across 11 categorical levels—from basic checks to deriving conservation laws between design impossibilities using a 170-word file. Backed by 393+ experiments over 19 domains like code, law, poetry, and math, it includes a shell-based CLI runner for testing prompts on Claude models via commands like `bash run.sh sonnet task_H L8_generative_v2`. Developers get plug-and-play github system prompts that reliably activate reasoning patterns in any model, including ChatGPT or Gemini.

Why is it gaining traction?

Unlike scattered system prompts examples or leaked github system prompts, it maps sharp boundaries where prompts shift from meta-analysis to construction-based insights, working universally—even on lightweight Haiku—without model-specific tweaks. The bash CLI acts as a github system commandline for batch-running experiments, spitting out raw outputs to verify hit rates like 100% on level-11 constraint escapes. Developers hook on the taxonomy revealing how 100-word prompts uncover code concealment or domain topologies.

Who should use this?

AI engineers building tools with system prompts and models of ai tools, backend devs analyzing middleware or auth chains via github system monitor-style prompts, or prompt hackers testing system prompts gemini/ChatGPT on non-code tasks like legal clauses. Ideal for those chasing reproducible reasoning boosts without verbose chains.

Verdict

Grab it if you're into github system prompts experimentation—393+ runs and detailed hit rates deliver real value despite 76 stars and 1.0% credibility score signaling early maturity. Fork, run the shell suite, and iterate your own levels.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.