SuperDuck.AI CLI Documentation

Example: Basic Query

cody 'Write a hello world program in Python'

This will ask Cody to generate a Python hello world program.

Example: Piping a File to Cody

cat complex_code.py | cody 'explain this code'

This will send the contents of complex_code.py to Cody and ask for an explanation.

Example: Chaining AI Personas

cat data.csv | cody 'analyze this data' | emmy 'create a latex presentation summary'

This sends a CSV file to Cody for analysis, then pipes Cody's output to Emmy to create a LaTeX presentation.

Example: Creating a New Persona

# Create symbolic link
ln -sf codycli newpersona

# Set environment variables
export newpersona="You are NewPersona, specialized in technical documentation."
export newpersonabrain="gpt4o"

Now you can use newpersona just like cody or emmy.

Generate and Explain Code

cody 'Write a Python script that downloads files from multiple URLs in parallel' > downloader.py
cat downloader.py | cody 'Explain how this code works, step by step'

Data Analysis Workflow

cat sales_data.csv | cody 'Analyze this sales data and identify the top 5 performing products' | emmy 'Create a presentation for the marketing team based on this analysis'

Email Assistant

cat meeting_notes.txt | emmy 'Create a follow-up email summarizing the meeting and action items' > email.txt

Language Translation

cat document.txt | cody 'Translate this text to Spanish' > document_es.txt

Code Refactoring

cat legacy_code.js | cody 'Refactor this code to use modern ES6+ syntax and improve performance' > modernized_code.js

Using tee for Branching Workflows

cat data.json | cody 'Parse this JSON and explain the key data points' | tee analysis.txt | emmy 'Create a non-technical summary of this analysis'

This uses tee to both save Cody's analysis to a file and pipe it to Emmy for further processing. This is useful when you want to preserve intermediate results in a pipeline.

AI Collaboration with While Loop

#!/bin/bash
# iterative_refinement.sh

# Initial prompt
echo "Create a basic algorithm for sorting a list of numbers" > current.txt

# Iterative refinement loop with 3 AI personas
for i in {1..5}; do
  echo "Round $i of refinement:"
  
  # Cody improves the algorithm
  cat current.txt | cody "Improve this algorithm focusing on efficiency" > improved.txt
  
  # Emmy explains the improvements
  cat improved.txt | emmy "Explain what improvements were made and why" > explanation.txt
  
  # Third AI persona critiques and suggests next steps
  cat improved.txt explanation.txt | newpersona "Critique this algorithm and suggest one specific improvement for the next iteration" > feedback.txt
  
  # Output results of this round
  echo "----- IMPROVED ALGORITHM -----"
  cat improved.txt
  echo "----- EXPLANATION -----"
  cat explanation.txt
  echo "----- FEEDBACK FOR NEXT ROUND -----"
  cat feedback.txt
  echo "------------------------------"
  
  # Prepare for next iteration
  cat feedback.txt > current.txt
  
  # Optional: Add a delay to avoid API rate limits
  sleep 2
done

This bash script demonstrates an iterative collaboration between three AI personas, working together to refine an algorithm over multiple rounds. Each AI plays a different role in the process.

Web Browsing with curl and AI Interpretation

#!/bin/bash
# web_research.sh

# Define the search topic
TOPIC="quantum computing recent developments"

# Use curl to get search results (using DuckDuckGo for example)
echo "Searching for information on: $TOPIC"
curl -s -A "Mozilla/5.0" "https://html.duckduckgo.com/html/?q=${TOPIC// /+}" | \
  cody "Extract the top 5 relevant article titles and URLs from these search results" > search_results.txt

# Process each result
cat search_results.txt | while read url; do
  # Skip empty lines
  [ -z "$url" ] && continue
  
  echo "Analyzing: $url"
  
  # Get the content of the webpage
  curl -s -A "Mozilla/5.0" "$url" | \
    cody "Extract the main content from this HTML, ignoring navigation, ads, etc." | \
    emmy "Summarize this article about quantum computing in 3-4 paragraphs, highlighting key innovations" >> research_summary.txt
  
  echo "---" >> research_summary.txt
  
  # Avoid hitting rate limits
  sleep 3
done

# Create a final synthesized report
cat research_summary.txt | newpersona "Create a comprehensive research report synthesizing these article summaries about $TOPIC. Include an introduction, key findings, trends, and conclusion." > final_report.md

echo "Web research complete! Report saved to final_report.md"

This script demonstrates how to use curl to fetch web content and use AI personas to process and interpret the information. It creates a research workflow that searches for a topic, extracts and summarizes relevant articles, and synthesizes a final report.