This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
- Run all tests with mods CLI:
./run_tests.sh - Run all tests with direct API calls:
./run_tests_curl.sh - Run with custom prompt:
./run_tests.sh -p "your prompt here"(or./run_tests_curl.sh -p "your prompt here") - Test specific models:
./run_tests.sh -m gpt-4o,sonnet-3.7(or./run_tests_curl.sh -m gpt-4o,sonnet-3.7) - Set experiment name:
./run_tests.sh -n experiment-name(or./run_tests_curl.sh -n experiment-name) - Deploy to Netlify:
./run_tests.sh -d(or./run_tests_curl.sh -d) - Show help:
./run_tests.sh -h(or./run_tests_curl.sh -h) - Install dependencies:
npm install - Set up API keys: Copy
.env.exampleto.envand add your keys (forrun_tests_curl.sh)
- Tests generate AI model implementations based on customizable prompts
- Two methods available: mods CLI (
run_tests.sh) or direct API calls (run_tests_curl.sh) - Results are stored in timestamped directories under
results/ - Each test run creates HTML outputs for each model tested
- Default settings are stored in
defaults.sh - API keys for direct calls should be stored in
.env(see.env.example) - Main index at results/index.html shows all experiments
- HTML/JS/CSS should be clean, self-contained, and ready to run
- Avoid markdown formatting in generated code
- Follow standard web development best practices
- Ensure generated code is properly formatted and indented
- When modifying shell scripts, follow POSIX-compatible syntax
- Use meaningful variable names and add comments where appropriate