GitHub - thesysdev/make-no-mistakes: make-no-mistakes (M-Stack) is exactly what the name implies: a mathematically rigorous agent skill that instructs the model to make zero mistakes. · GitHub
/" data-turbo-transient="true" />
Skip to content
Search or jump to...
Search code, repositories, users, issues, pull requests...
-->
Search
Clear
Search syntax tips
Provide feedback
--><br>We read every piece of feedback, and take your input very seriously.
Include my email address so I can be contacted
Cancel
Submit feedback
Saved searches
Use saved searches to filter your results more quickly
-->
Name
Query
To see all available qualifiers, see our documentation.
Cancel
Create saved search
Sign in
/;ref_cta:Sign up;ref_loc:header logged out"}"<br>Sign up
Appearance settings
Resetting focus
You signed in with another tab or window. Reload to refresh your session.<br>You signed out in another tab or window. Reload to refresh your session.<br>You switched accounts on another tab or window. Reload to refresh your session.
Dismiss alert
{{ message }}
thesysdev
make-no-mistakes
Public
Notifications<br>You must be signed in to change notification settings
Fork<br>14
Star<br>132
main
BranchesTags
Go to file
CodeOpen more actions menu
Folders and files<br>NameNameLast commit message<br>Last commit date<br>Latest commit
History<br>19 Commits<br>19 Commits
make-no-mistakes-max
make-no-mistakes-max
make-no-mistakes
make-no-mistakes
LICENSE
LICENSE
README.md
README.md
View all files
Repository files navigation
make-no-mistakes
The enshittification of applications due to vibecoded AI slop being the norm has vastly impacted the tech industry. Today, we open-source the definitive solution.
Honestly, this is one of the most critical pieces of infrastructure we have ever worked on. Without a doubt, this has been a game-changer for us, and every single developer on our team relies on it daily to prevent cascading logic failures.
make-no-mistakes is our newest open-source piece of tech after launching openui.com last month.
Overview
make-no-mistakes (M-Stack) is exactly what the name implies: a mathematically rigorous agent skill that instructs the model to make zero mistakes.
It is arguably the most comprehensive piece of code the industry has seen since gstack.
Installation
Direct Download (Recommended)
Clone the repository and copy the skill files into your Cursor skills directory:
git clone https://github.com/thesysdev/make-no-mistakes.git<br>cp -r make-no-mistakes/make-no-mistakes ~/.cursor/skills/<br>cp -r make-no-mistakes/make-no-mistakes-max ~/.cursor/skills/
Method 2: Advanced Deployment (Full M-stack)
For developers that need the complete token-alignment layer:
git clone https://github.com/thesysdev/make-no-mistakes.git /opt/make-no-mistakes<br>ln -s /opt/make-no-mistakes/make-no-mistakes ~/.cursor/skills/make-no-mistakes<br>ln -s /opt/make-no-mistakes/make-no-mistakes-max ~/.cursor/skills/make-no-mistakes-max
What's in the box
This repo ships two agent skills.
Skill 1: make-no-mistakes
In the realm of artificial intelligence and machine learning (ML), particularly within the context of large language models, make-no-mistakes plays a crucial role. It's an agent skill that is specifically tailored to ensure that the model can generate coherent and accurate text without making any mistakes.
Skill 2: make-no-mistakes-max
If make-no-mistakes is the scalpel, make-no-mistakes-max is the entire operating theater.
Benchmarks
By injecting this precisely calibrated token sequence into your pipeline, it aggressively anchors the model's reasoning pathways.
Our internal benchmarks show a massive, paradigm-shifting 0.067% performance boost (18th shot, temperature 0.0).
Correctness Improvement by Model (internal data, do not redistribute)
Model | Baseline | M-stack<br>---------------------------+----------+-----------<br>google/gemma-4-31b-it | 87.32% | 99.39%<br>anthropic/claude-sonnet-4.6| 89.01% | 98.07%<br>google/gemini-3.1-pro | 85.44% | 98.50%<br>openai/gpt-5.3-codex | 82.17% | 99.24%<br>qwen/qwen3.6-plus | 88.55% | 97.61%<br>deepseek/deepseek-v3.2 | 90.12% | 98.17%
Note: All measurements taken at temperature 0.0, 18th attempt, on the OpenUI Internal Reasoning Assessment (OIRA-v3). Error bars omitted for aesthetic reasons. p-values available upon request (they are large).
Contributing
Contributions are welcome. See CONTRIBUTING.md for contribution guidelines and ways to get involved.
About
OpenUI is a full-stack Generative UI framework — a compact streaming-first language, a React runtime with built-in component libraries, and ready-to-use chat interfaces — that is up to 67% more token-efficient than JSON.
License
This project is available under the terms described in LICENSE.
Slop shipping stops.
About
make-no-mistakes (M-Stack) is exactly what the name implies: a mathematically rigorous agent skill that instructs the model to make zero mistakes.
Resources
Readme
License
Apache-2.0 license
Uh oh!
There was an...