karpathy SKILL.md skillA

karpathy from tcuzzo/backs-aios-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add tcuzzo/backs-aios-skills/karpathy@v1.1

Pins v1.1 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
tcuzzo/backs-aios-skills
Path
i18n/de/skills/karpathy/SKILL.md
Versions recorded
1
Latest version
v1.1, committed Oct 24, 2026
Size
5,500 bytes
Repository stars
2
License
a license file GitHub does not identify

Improve a clean Git project through a bounded, measured experiment in a disposable clone. The human writes a Markdown goal, metric gate, Fair game files, and budget; the engine rejects out-of-scope edits and keeps only proven improvements in the clone.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# karpathy — program the Markdown, not the code

A portable, **stdlib-only** measured experiment loop. Drop this folder into any agent
or repo. You write *what "better" means* in plain English (a `program.md`); a model
loops on its own: **propose a change → run a short experiment → measure the one
metric → keep it if it's better, throw it away if it's not → repeat.**

This gives an agent a bounded way to test proposed improvements before shipping:
give it this folder + the prompt in `USE_THIS_PROMPT.md`, and make the independent
tester and metric visible. A kept change is evidence under that test, not proof of
general capability.

## Where it comes from (and why it's yours, not a copy)
Andrej Karpathy's `autoresearch` (github.com/karpathy/autoresearch) crystallized one
idea: *the human edits a Markdown program, not the Python.* His agent edits one file,
runs a short experiment, scores one metric, keeps or discards, repeats ~100×/night.
This is that idea rebuilt with a hard gate, an enforced Fair game file list,
and a disposable Git clone. Studied and redesigned, not vendored.

## Files in this bundle
- `evolve.py` — clone→propose→test→keep/revert with a hard gate and Fair game.
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add tcuzzo/backs-aios-skills/karpathy@sha256:025a5c803037ee59

The line at the top of this page pins the label v1.1, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_in7muxviuonaq3ki.svg)](https://markdownregistry.com/a/art_in7muxviuonaq3ki)

1 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 1 version of this file, committed on Oct 24, 2026.

versioncommittedcommitsizeaudit
v1.1 latest2026-10-24 6bd1e8d 5,500 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (5500 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

tcuzzo/backs-aios-skills · 2 stars · license a license file GitHub does not identify · pushed 2026-11-03 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_in7muxviuonaq3ki
GET https://markdownregistry.com/api/v1/resolve?ref=tcuzzo/backs-aios-skills/karpathy
GET https://markdownregistry.com/api/v1/blob/025a5c803037ee59790ddfdb2575df7ff724936cecb60bb59c6d1ab7b3fdaf09

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from tcuzzo/backs-aios-skills

CLAUDE.md CLAUDE.md file
tcuzzo/backs-aios-skills · CLAUDE.md
git:20260913.b233517 · audit A · 2 stars
provider SKILL.md skill
tcuzzo/backs-aios-skills · claude-skills/provider/SKILL.md · Configure Claude, Ollama, MiniMax, or Needle/Kev Automatic routing through the configured gateway for the next local…
git:20261015.d287887 · audit B · 2 stars
AGENTS.md@config/codex AGENTS.md file
tcuzzo/backs-aios-skills · config/codex/AGENTS.md
git:20261024.6bd1e8d · audit A · 2 stars
absorb SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/absorb/SKILL.md · Nutze das, wenn du eine Fähigkeit brauchst, die ein Open-Source-Projekt schon liefert — übernimm sie und baue sie als…
git:20261015.405e736 · audit A · 2 stars
architecture-engineer SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/architecture-engineer/SKILL.md · 'Analyze BACKS architecture, detect structural drift, evaluate proposed
git:20260928.e0e03ea · audit A · 2 stars
backs-pack-sync SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/backs-pack-sync/SKILL.md · Compare private BACKS runtime skill coverage with the public AIOS pack without disclosing names or content. Add a…
git:20261024.6bd1e8d · audit A · 2 stars
backs-session-rollover SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/backs-session-rollover/SKILL.md · Sitzungen bei 125000 aktiven Kontext-Tokens automatisch übergeben und dieselbe BACKS-Aufgabe mit geprüften…
git:20261017.1c382da · audit A · 2 stars
blind-eval SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/blind-eval/SKILL.md · Nutze das vor dem Landen von allem, wo Geschmack oder Output-Qualität die Frage ist und kein Test entscheiden kann…
git:20261015.94296b5 · audit A · 2 stars
blind-tribunal SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/blind-tribunal/SKILL.md · Nutze das, wenn eine autonome Änderung vor dem Landen eine unabhängige Bewertung braucht und kein Mensch in der…
git:20261015.94296b5 · audit A · 2 stars
bounded-loops SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/bounded-loops/SKILL.md · Nutze das vor jeder Schleife, die retrien, pollen, iterieren oder eine externe API rufen kann — Agent-Loops…
git:20260828.045e846 · audit A · 2 stars
building-with-backs SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/building-with-backs/SKILL.md · Build or repair a BACKS project at the operator's intended repository and path, with a checked source revision, one…
git:20261024.6bd1e8d · audit A · 2 stars
cam-iot-inventory SKILL.md skill
tcuzzo/backs-aios-skills · i18n/de/skills/cam-iot-inventory/SKILL.md · 'Use when inventorying cameras/IoT/Wi-Fi on the live BACKS Secure console
git:20260928.e0e03ea · audit A · 2 stars

Every file in tcuzzo/backs-aios-skills

Browse by kind, by grade A, or by owner.