Anubis
Developers often try multiple models manually, compare code quality by eye, then hope the one they picked works.
Core trick
Fans one prompt out to several models in parallel, then ranks candidate code by a user-weighted score across quality metrics.
Tags
Description
Developers often try multiple models manually, compare code quality by eye, then hope the one they picked works. This is slow and inconsistent.
Anubis takes a single prompt and automatically runs it through multiple AI models. Each model generates a code candidate.
Anubis then automatically evaluates each candidate using a scoring pipeline based on user-selected priorities. Each metric is weighted according to the priority level chosen by the user. Anubis produces a final weighted score for each code generated, ranks them, and surfaces the best one.
README excerpt
# Anubis ⚖️𓁣 **Automated AI Code Comparison & Evaluation System** [](https://github.com/w3joe/anubis) ## 📖 Project Description Anubis is an intelligent code evaluation platform that automates the process of comparing code generation across multiple AI models. Instead of manually testing different models and hoping for the best, Anubis takes a single prompt and automatically runs it through multiple AI models in parallel, evaluates each candidate using a comprehensive scoring pipeline, and surfaces the best result based on your priorities. ## 🎯 The Problem Developers often face a frustrating workflow when using AI code generation: 1. **Manual Testing**: Developers try multiple AI models manually, one at a time 2. **Subjective Comparison**: Code quality is judged by eye, leading to inconsistent decisions 3. **Uncertainty**: No objective metrics to determine which model produces the best code for a specific task 4. **No Priority System**: Can't easily prioritize what matters most (e.g., performance vs. readability) This process is slow, inconsistent, and doesn't scale. ## ✨ The Solution Anubis automates the entire workflow: 1. **Parallel Multi-Model Generation**: Takes a single prompt and runs it through multiple AI models simultaneously in different branches 2. **Automated Evaluation**: Each code candidate is automatically evaluated using a comprehensive scoring pipeline 3. **Priority-Based Weighting**