Common Workflows

This guide presents real-world usage patterns and practical examples for the kagi CLI. Each workflow is designed to solve a specific problem or automate a common task, with complete copy-pasteable examples.

Workflow Categories

Daily Automation - News briefings, routine searches
Research & Investigation - Deep research workflows
Content Processing - Summarization and extraction
Development - Programming and debugging assistance
AI Integration - Assistant and FastGPT patterns
Data Pipelines - Processing multiple items

Daily Automation Workflows

Morning News Briefing

Start your day with a curated news summary:

#!/bin/bash
# ~/bin/morning-brief.sh

echo "🌅 Morning Brief - $(date '+%Y-%m-%d %H:%M')"
echo "================================================"
echo ""

echo "📰 Technology"
echo "-------------"
kagi news --category tech --limit 5 | jq -r '.stories[] | "• \(.title)\n  \(.articles[0].link)\n"'

echo ""
echo "🌍 World News"
echo "-------------"
kagi news --category world --limit 5 | jq -r '.stories[] | "• \(.title)\n  \(.articles[0].link)\n"'

echo ""
echo "💼 Business"
echo "------------"
kagi news --category business --limit 3 | jq -r '.stories[] | "• \(.title)\n  \(.articles[0].link)\n"'

echo ""
echo "🔬 Science"
echo "-----------"
kagi news --category science --limit 3 | jq -r '.stories[] | "• \(.title)\n  \(.articles[0].link)\n"'

Usage:

chmod +x ~/bin/morning-brief.sh
~/bin/morning-brief.sh

Add to cron for automated delivery:

# Edit crontab
crontab -e

# Add line for 8 AM daily
0 8 * * * ~/bin/morning-brief.sh >> ~/news-brief.log 2>&1

Weekly Research Digest

Collect and summarize interesting articles:

#!/bin/bash
# ~/bin/weekly-digest.sh

WEEK_DIR="$HOME/research/$(date +%Y-week-%U)"
mkdir -p "$WEEK_DIR"

# Small Web discoveries
echo "Saving Small Web feed..."
kagi smallweb --limit 20 > "$WEEK_DIR/smallweb.json"

# Extract URLs and summarize top 5
jq -r '.data[0:5] | .[].url' "$WEEK_DIR/smallweb.json" | while read -r url; do
  echo "Summarizing: $url"
  filename=$(echo "$url" | sed 's/[^a-zA-Z0-9]/_/g').json
  kagi summarize --subscriber --url "$url" --length overview > "$WEEK_DIR/$filename" 2>/dev/null || echo "Failed: $url"
  sleep 2  # Be nice to the API
done

echo "Digest saved to: $WEEK_DIR"

Daily Standup Helper

Prepare answers for common standup questions:

#!/bin/bash
# ~/bin/standup-prep.sh

YESTERDAY=$(date -d "yesterday" +%Y-%m-%d)

echo "📝 Standup Prep for $(date '+%Y-%m-%d')"
echo "======================================"
echo ""

echo "Yesterday ($YESTERDAY):"
echo "----------------------"
# Search your notes or activity
# Example: search for commits, PRs, tickets

echo ""
echo "Today:"
echo "------"
echo "- [ ] ..."
echo "- [ ] ..."

echo ""
echo "Blockers:"
echo "---------"
echo "- None / ..."

echo ""
echo "📚 Quick Research:"
echo "------------------"
# Example: get latest on your tech stack
kagi news --category tech --limit 3 | jq -r '.stories[] | "• \(.title)"'

Research & Investigation Workflows

Deep Topic Research

Systematically explore a topic across multiple dimensions:

#!/bin/bash
# ~/bin/research-topic.sh

TOPIC="$1"
if [ -z "$TOPIC" ]; then
  echo "Usage: research-topic.sh 'your research topic'"
  exit 1
fi

RESEARCH_DIR="$HOME/research/$(echo "$TOPIC" | tr ' ' '_' | tr '[:upper:]' '[:lower:]')-$(date +%Y%m%d)"
mkdir -p "$RESEARCH_DIR"

echo "🔍 Researching: $TOPIC"
echo "======================"
echo ""

# 1. Initial search
echo "Step 1: General search..."
kagi search --format pretty "$TOPIC" > "$RESEARCH_DIR/01-general-search.txt"

# 2. News coverage
echo "Step 2: Recent news..."
kagi news --limit 10 | jq -r '.stories[] | select(.title | contains("'$TOPIC'")) | "\(.title)\n\(.articles[0].link)\n"' > "$RESEARCH_DIR/02-news.txt"

# 3. FastGPT overview
echo "Step 3: Getting overview..."
kagi fastgpt "Provide a comprehensive overview of $TOPIC" > "$RESEARCH_DIR/03-overview.txt"

# 4. Web enrichment for deeper insights
echo "Step 4: Deep web search..."
kagi enrich web "$TOPIC" > "$RESEARCH_DIR/04-web-enrichment.json"

# 5. Save top 5 results for later reading
echo "Step 5: Saving top results..."
kagi search "$TOPIC" | jq -r '.data[0:5] | .[].url' > "$RESEARCH_DIR/05-urls-to-read.txt"

echo ""
echo "✅ Research complete!"
echo "📁 Saved to: $RESEARCH_DIR"
echo ""
echo "Next steps:"
echo "  - Review $RESEARCH_DIR/03-overview.txt"
echo "  - Read URLs in $RESEARCH_DIR/05-urls-to-read.txt"
echo "  - Summarize key articles with kagi summarize"

Competitor Analysis

Research companies or products systematically:

#!/bin/bash
# ~/bin/competitor-analysis.sh

COMPANY="$1"
RESEARCH_DIR="$HOME/competitor-research/$(echo "$COMPANY" | tr ' ' '_')-$(date +%Y%m%d)"
mkdir -p "$RESEARCH_DIR"

echo "🏢 Competitor Analysis: $COMPANY"
echo "================================="
echo ""

# Company overview
echo "1. Company Overview"
kagi fastgpt "What does $COMPANY do? What are their main products and services?" > "$RESEARCH_DIR/01-overview.txt"

# Recent news
echo "2. Recent News"
kagi search --format pretty "$COMPANY news $(date +%Y)" > "$RESEARCH_DIR/02-recent-news.txt"

# Product comparison
echo "3. Product Information"
kagi search --format pretty "$COMPANY products features" > "$RESEARCH_DIR/03-products.txt"

# Market position
echo "4. Market Position"
kagi fastgpt "What is $COMPANY's market position? Who are their main competitors?" > "$RESEARCH_DIR/04-market-position.txt"

# Technical analysis (if applicable)
echo "5. Technical Stack"
kagi search --format pretty "$COMPANY technology stack infrastructure" > "$RESEARCH_DIR/05-technology.txt"

echo ""
echo "Analysis complete! Check $RESEARCH_DIR"

Academic Research Assistant

#!/bin/bash
# ~/bin/academic-research.sh

QUERY="$1"
RESEARCH_DIR="$HOME/academic-research/$(date +%Y%m%d-%H%M%S)"
mkdir -p "$RESEARCH_DIR"

echo "📚 Academic Research: $QUERY"
echo "============================="
echo ""

# Search for papers and resources
echo "Searching academic resources..."
kagi search --format pretty "$QUERY filetype:pdf OR site:arxiv.org OR site:scholar.google.com" > "$RESEARCH_DIR/academic-sources.txt"

# Get summary
echo "Generating summary..."
kagi fastgpt "Summarize the current state of research on: $QUERY" > "$RESEARCH_DIR/summary.txt"

# Find key papers
echo "Identifying key papers..."
kagi search "$QUERY most cited papers" | jq -r '.data[0:5] | .[].url' > "$RESEARCH_DIR/key-papers.txt"

echo "Saved to: $RESEARCH_DIR"

Content Processing Workflows

URL Summarization Pipeline

Process a list of URLs and generate summaries:

#!/bin/bash
# ~/bin/summarize-urls.sh

URL_FILE="$1"
OUTPUT_DIR="${2:-./summaries}"

if [ ! -f "$URL_FILE" ]; then
  echo "Usage: summarize-urls.sh <urls-file> [output-directory]"
  echo "File should contain one URL per line"
  exit 1
fi

mkdir -p "$OUTPUT_DIR"

echo "📝 Processing URLs from: $URL_FILE"
echo "Output directory: $OUTPUT_DIR"
echo ""

line_num=0
while IFS= read -r url; do
  line_num=$((line_num + 1))
  
  # Skip empty lines and comments
  [[ -z "$url" || "$url" =~ ^# ]] && continue
  
  echo "[$line_num] Processing: $url"
  
  # Create safe filename
  filename=$(echo "$url" | sed 's|https*://||; s|[^a-zA-Z0-9]|_|g; s|_$||').json
  output_path="$OUTPUT_DIR/$filename"
  
  # Summarize with subscriber mode
  if kagi summarize --subscriber --url "$url" --length overview --summary-type keypoints > "$output_path" 2>/dev/null; then
    echo "  ✓ Saved to: $filename"
  else
    echo "  ✗ Failed to summarize"
    echo "$url" >> "$OUTPUT_DIR/failed.txt"
  fi
  
  # Rate limiting - be nice to the service
  sleep 1
done < "$URL_FILE"

echo ""
echo "✅ Complete!"
echo "Summaries saved to: $OUTPUT_DIR"
[ -f "$OUTPUT_DIR/failed.txt" ] && echo "Failed URLs: $OUTPUT_DIR/failed.txt"

Usage:

# Create URL list
cat > urls.txt << 'EOF'
https://example.com/article1
https://example.com/article2
https://example.com/article3
EOF

# Run pipeline
summarize-urls.sh urls.txt ./my-summaries

Content Curation

Build a curated reading list with summaries:

#!/bin/bash
# ~/bin/curate-reading-list.sh

CATEGORY="$1"
LIMIT="${2:-10}"

echo "📖 Curating Reading List: $CATEGORY"
echo "===================================="
echo ""

# Search for recent content
echo "Finding recent articles..."
kagi search --format pretty "$CATEGORY $(date +%Y) guide tutorial" > /tmp/search-results.txt

# Extract top URLs and summarize
echo ""
echo "Summarizing top $LIMIT articles..."
kagi search "$CATEGORY $(date +%Y)" | jq -r ".data[0:$LIMIT] | .[].url" | while read -r url; do
  echo ""
  echo "Article: $url"
  echo "---"
  kagi summarize --subscriber --url "$url" --length headline 2>/dev/null | jq -r '.data.output' || echo "Could not summarize"
  echo ""
  sleep 1
done

Development Workflows

Error Investigation

Research error messages and find solutions:

#!/bin/bash
# ~/bin/debug-error.sh

ERROR_MSG="$1"
CONTEXT="${2:-}"

echo "🐛 Debugging Error"
echo "=================="
echo "Error: $ERROR_MSG"
[ -n "$CONTEXT" ] && echo "Context: $CONTEXT"
echo ""

# Search for the error
echo "🔍 Searching for solutions..."
kagi search --format pretty "\"$ERROR_MSG\" solution fix"

echo ""
echo "🤖 AI Analysis..."
FULL_CONTEXT="Error: $ERROR_MSG"
[ -n "$CONTEXT" ] && FULL_CONTEXT="$FULL_CONTEXT. Context: $CONTEXT"
kagi assistant "I'm getting this error in my code. Please help me understand what's causing it and how to fix it: $FULL_CONTEXT"

Usage:

debug-error.sh "TypeError: Cannot read property 'map' of undefined" "React component"

Technology Evaluation

Research technologies before adopting:

#!/bin/bash
# ~/bin/evaluate-tech.sh

TECH="$1"

echo "🔬 Technology Evaluation: $TECH"
echo "================================"
echo ""

# Overview
echo "1. Overview"
kagi fastgpt "What is $TECH and what problem does it solve?"

# Pros and cons
echo ""
echo "2. Pros and Cons"
kagi fastgpt "What are the main advantages and disadvantages of $TECH compared to alternatives?"

# Community and ecosystem
echo ""
echo "3. Community"
kagi search --format pretty "$TECH github stars community adoption $(date +%Y)"

# Recent developments
echo ""
echo "4. Recent News"
kagi news --category tech --limit 5 | jq -r '.stories[] | select(.title | contains("'$TECH'")) | "\(.title)"'

# Learning resources
echo ""
echo "5. Learning Resources"
kagi search --format pretty "$TECH tutorial getting started documentation"

API Documentation Search

Find relevant documentation quickly:

#!/bin/bash
# ~/bin/api-docs.sh

API="$1"
QUERY="$2"

echo "📖 $API Documentation Search"
echo "============================"
echo "Query: $QUERY"
echo ""

# Search specific documentation
case "$API" in
  "react")
    kagi search --format pretty "site:react.dev $QUERY"
    ;;
  "node")
    kagi search --format pretty "site:nodejs.org $QUERY"
    ;;
  "mdn")
    kagi search --format pretty "site:developer.mozilla.org $QUERY"
    ;;
  *)
    kagi search --format pretty "$API documentation $QUERY"
    ;;
esac

AI Integration Workflows

Conversation Chains

Build on previous Assistant responses:

#!/bin/bash
# ~/bin/chain-conversation.sh

# Initial prompt
THREAD_ID=$(kagi assistant "Explain the concept of recursion in programming" | jq -r '.thread.id')
echo "Started thread: $THREAD_ID"

# Follow-up 1
kagi assistant --thread-id "$THREAD_ID" "Can you provide a practical example in Python?"

# Follow-up 2
kagi assistant --thread-id "$THREAD_ID" "Now show me how to avoid stack overflow in deep recursion"

# Follow-up 3
kagi assistant --thread-id "$THREAD_ID" "Compare this to an iterative approach"

echo ""
echo "To continue this conversation later, use:"
echo "kagi assistant --thread-id $THREAD_ID \"your question\""

FastGPT Decision Support

Use FastGPT for quick technical decisions:

#!/bin/bash
# ~/bin/tech-decision.sh

QUESTION="$1"

echo "🤔 Decision Support"
echo "==================="
echo "Question: $QUESTION"
echo ""

# Get quick analysis
kagi fastgpt "$QUESTION"

echo ""
echo "For deeper analysis, try:"
echo "kagi assistant \"Provide a detailed analysis of: $QUESTION\""

Content Generation

Generate content with AI assistance:

#!/bin/bash
# ~/bin/generate-content.sh

TYPE="$1"  # email, doc, summary
TOPIC="$2"

echo "✍️  Content Generation: $TYPE"
echo "Topic: $TOPIC"
echo ""

case "$TYPE" in
  "email")
    kagi assistant "Write a professional email about: $TOPIC"
    ;;
  "doc")
    kagi assistant "Write technical documentation for: $TOPIC"
    ;;
  "summary")
    kagi assistant "Create an executive summary of: $TOPIC"
    ;;
  *)
    kagi assistant "Write about: $TOPIC"
    ;;
esac

Data Pipeline Workflows

Batch URL Processing

Process multiple URLs with error handling:

#!/bin/bash
# ~/bin/batch-process.sh

INPUT_FILE="$1"
PROCESS_TYPE="${2:-summarize}"  # summarize, check, extract

if [ ! -f "$INPUT_FILE" ]; then
  echo "Usage: batch-process.sh <input-file> [process-type]"
  exit 1
fi

OUTPUT_DIR="batch-output-$(date +%Y%m%d-%H%M%S)"
mkdir -p "$OUTPUT_DIR"

SUCCESS=0
FAILED=0

while IFS= read -r line; do
  [ -z "$line" ] && continue
  
  case "$PROCESS_TYPE" in
    "summarize")
      if kagi summarize --subscriber --url "$line" > "$OUTPUT_DIR/$(echo "$line" | md5sum | cut -d' ' -f1).json" 2>/dev/null; then
        ((SUCCESS++))
        echo "✓ $line"
      else
        ((FAILED++))
        echo "✗ $line"
        echo "$line" >> "$OUTPUT_DIR/failed.txt"
      fi
      ;;
    "check")
      # Check if URL is accessible
      if curl -s -o /dev/null -w "%{http_code}" "$line" | grep -q "200"; then
        echo "✓ $line (200 OK)"
        ((SUCCESS++))
      else
        echo "✗ $line (failed)"
        ((FAILED++))
      fi
      ;;
    "extract")
      # Extract just the titles
      kagi summarize --subscriber --url "$line" --length headline 2>/dev/null | jq -r '.data.output' >> "$OUTPUT_DIR/titles.txt"
      ((SUCCESS++))
      ;;
  esac
  
  sleep 1
done < "$INPUT_FILE"

echo ""
echo "Results:"
echo "  Success: $SUCCESS"
echo "  Failed: $FAILED"
echo "  Output: $OUTPUT_DIR"

JSON Data Pipeline

Process search results programmatically:

#!/bin/bash
# ~/bin/search-pipeline.sh

QUERY="$1"

echo "🔍 Search Pipeline: $QUERY"
echo "==========================="
echo ""

# Search and process
kagi search "$QUERY" | jq -r '
  .data |
  group_by(.t) |
  map({
    type: (.[0].t | tostring),
    count: length,
    urls: [.[].url]
  }) |
  .[] |
  "Type \(.type): \(.count) results"
'

# Extract just organic results
kagi search "$QUERY" | jq -r '
  .data |
  map(select(.t == 0)) |
  .[0:5] |
  .[] |
  "\(.title)\n\(.url)\n\(.snippet)\n---"
'

Workflow Best Practices

Rate Limiting

Always include delays when processing multiple items:

# Good - includes delay
for url in $URLS; do
  kagi summarize --subscriber --url "$url"
  sleep 1  # Be respectful
done

# Bad - hammers the API
for url in $URLS; do
  kagi summarize --subscriber --url "$url"
done

Error Handling

Always handle failures gracefully:

# Good - handles errors
if kagi search "$query" > result.json 2>/dev/null; then
  echo "Success"
else
  echo "Failed to search"
fi

# With retry logic
for i in 1 2 3; do
  if kagi search "$query" > result.json 2>/dev/null; then
    break
  fi
  sleep 5
done

Output Management

Organize outputs systematically:

# Create dated directories
OUTPUT_DIR="$HOME/kagi-output/$(date +%Y/%m/%d)"
mkdir -p "$OUTPUT_DIR"

# Use descriptive filenames
kagi search "topic" > "$OUTPUT_DIR/search-topic-$(date +%H%M%S).json"

# Log operations
exec 1> >(tee -a "$OUTPUT_DIR/log.txt")
exec 2>&1

Configuration Management

Use environment variables for flexibility:

#!/bin/bash
# Configurable script

RESULTS_LIMIT="${KAGI_RESULTS_LIMIT:-10}"
SUMMARY_LENGTH="${KAGI_SUMMARY_LENGTH:-overview}"
OUTPUT_FORMAT="${KAGI_OUTPUT_FORMAT:-json}"

kagi search "$QUERY" | jq ".data[0:${RESULTS_LIMIT}]"

Advanced Patterns

Combining Commands

Chain kagi with other tools:

# Search and open first result in browser
kagi search "$QUERY" | jq -r '.data[0].url' | xargs open

# Search and add to todo list
kagi search "$QUERY" | jq -r '.data[] | "- [ ] \(.title) - \(.url)"' >> todo.md

# Generate report with template
cat template.md | sed "s/{{QUERY}}/$QUERY/g" | kagi assistant "Fill in this template based on: $QUERY"

Conditional Workflows

Adapt based on available tokens:

#!/bin/bash

# Check auth and adapt behavior
if kagi auth check 2>/dev/null; then
  # Full features available
  kagi search --lens 2 "$QUERY"
else
  # Fallback to public commands
  kagi news --limit 5
  echo "Note: Set up authentication for full search capabilities"
fi

Parallel Processing

Process independent items in parallel:

#!/bin/bash

# Process URLs in parallel (be careful with rate limits!)
cat urls.txt | xargs -P 3 -I {} bash -c '
  kagi summarize --subscriber --url "$1" > "summaries/$(echo "$1" | md5sum | cut -d" " -f1).json" 2>/dev/null
' _ {}

Note: Be very careful with parallelism to avoid rate limiting.

Workflow Templates

Template 1: Research Project

#!/bin/bash
PROJECT="$1"
DIR="$HOME/research-projects/$PROJECT"
mkdir -p "$DIR"

cd "$DIR" || exit

# Initialize
git init 2>/dev/null || true

# Research phase
kagi search "$PROJECT overview" > 01-overview.json
kagi news "$PROJECT" > 02-news.json

# Analysis phase
kagi fastgpt "Analyze $PROJECT" > 03-analysis.txt

# Documentation
cat > README.md << EOF
# Research: $PROJECT

Date: $(date)

## Overview
See 01-overview.json

## News
See 02-news.json

## Analysis
See 03-analysis.txt
EOF

echo "Research project initialized in $DIR"

Template 2: Weekly Review

#!/bin/bash
WEEK="$(date +%Y-week-%U)"
DIR="$HOME/weekly-reviews/$WEEK"
mkdir -p "$DIR"

echo "# Weekly Review: $WEEK" > "$DIR/review.md"
echo "" >> "$DIR/review.md"
echo "## Technology News" >> "$DIR/review.md"
kagi news --category tech --limit 10 >> "$DIR/review.md"

echo "" >> "$DIR/review.md"
echo "## Small Web Highlights" >> "$DIR/review.md"
kagi smallweb --limit 10 >> "$DIR/review.md"

echo "Weekly review saved to $DIR/review.md"

Next Steps

Advanced Usage - Scripting, CI/CD, automation
Command Reference - Detailed command documentation
Troubleshooting - Debugging workflows

Create your own workflows and share them with the community!

Getting Started

Usage

​Common Workflows

​Workflow Categories

​Daily Automation Workflows

​Morning News Briefing

​Weekly Research Digest

​Daily Standup Helper

​Research & Investigation Workflows

​Deep Topic Research

​Competitor Analysis

​Academic Research Assistant

​Content Processing Workflows

​URL Summarization Pipeline

​Content Curation

​Development Workflows

​Error Investigation

​Technology Evaluation

​API Documentation Search

​AI Integration Workflows

​Conversation Chains

​FastGPT Decision Support

​Content Generation

​Data Pipeline Workflows

​Batch URL Processing

​JSON Data Pipeline

​Workflow Best Practices

​Rate Limiting

​Error Handling

​Output Management

​Configuration Management

​Advanced Patterns

​Combining Commands

​Conditional Workflows

​Parallel Processing

​Workflow Templates

​Template 1: Research Project

​Template 2: Weekly Review

​Next Steps

Common Workflows

Workflow Categories

Daily Automation Workflows

Morning News Briefing

Weekly Research Digest

Daily Standup Helper

Research & Investigation Workflows

Deep Topic Research

Competitor Analysis

Academic Research Assistant

Content Processing Workflows

URL Summarization Pipeline

Content Curation

Development Workflows

Error Investigation

Technology Evaluation

API Documentation Search

AI Integration Workflows

Conversation Chains

FastGPT Decision Support

Content Generation

Data Pipeline Workflows

Batch URL Processing

JSON Data Pipeline

Workflow Best Practices

Rate Limiting

Error Handling

Output Management

Configuration Management

Advanced Patterns

Combining Commands

Conditional Workflows

Parallel Processing

Workflow Templates

Template 1: Research Project

Template 2: Weekly Review

Next Steps