pr-authorship-analysis

name	pr-authorship-analysis
description	Analyze PR authorship from the prior calendar month, identifying real human authors behind bot-created PRs. Use when users want monthly PR analysis, contributor statistics, or team productivity reports.

Monthly PR Authorship Analysis Skill

This skill provides comprehensive analysis of all pull requests merged in the prior calendar month, with special focus on identifying the real human contributors behind bot-created PRs by examining commit histories.

When to Use This Skill

Use this skill when users ask about:

• Monthly PR analysis and contribution reports
• "PRs from last month" or "prior month's PRs"
• Monthly team productivity statistics
• Finding real contributors behind bot PRs from previous month
• Monthly contributor rankings
• Team performance for the completed month
• "Who worked on what last month" questions

Core Methodology

1. Prior Month Date Calculation

Automatic Date Range:

# Calculate prior month start and end dates
CURRENT_DATE=$(date +%Y-%m-%d)
PRIOR_MONTH_START=$(date -d "$(date +%Y-%m-01) -1 month" +%Y-%m-01)
PRIOR_MONTH_END=$(date -d "$(date +%Y-%m-01) -1 day" +%Y-%m-%d)

echo "Analyzing PRs from: $PRIOR_MONTH_START to $PRIOR_MONTH_END"

Complete Data Collection:

# Get accurate total count first
gh pr list --state merged --search "merged:>=$PRIOR_MONTH_START merged:<=$PRIOR_MONTH_END" --json number --limit 1000 | jq 'length'

# Get complete PR dataset for prior month
gh pr list --state merged --search "merged:>=$PRIOR_MONTH_START merged:<=$PRIOR_MONTH_END" --json number,title,author,mergedAt,body --limit 1000

Important: Always verify the actual count vs requested limit. Round numbers (like 100, 500) often indicate you hit the limit rather than the true count.

2. Bot PR Identification

Look for PRs authored by common bot accounts:

• app/kilo-code-bot
• kilo-code-bot[bot]
• kiloconnect[bot]
• Any account with [bot] suffix

3. Human Author Discovery

For each bot PR, analyze commit history:

gh pr view [PR_NUMBER] --json commits

Commit Analysis Pattern:

• Extract authors array from each commit
• Identify human vs bot authors by:
  • Email domains (avoid @users.noreply.github.com patterns for bots)
  • Login names (avoid [bot] suffixes)
  • Real human names vs system names

Attribution Rules:

• Human + Bot Collaboration: Most common - bot creates initial commits, human refines
• Pure Human: Fully human work misclassified as bot PR
• Pure Bot: Genuinely automated with no human commits

4. Statistical Analysis

Author Mapping:

• Direct PRs (straightforward human authorship)
• Bot PRs → Human contributors (via commit analysis)
• Combined totals for accurate attribution

Key Metrics to Track:

• Total PRs analyzed
• Bot vs Human PR counts
• Human authors found in bot PRs vs purely automated
• Top contributors (direct + bot PR contributions)
• Collaboration patterns

Implementation Workflow

Phase 1: Prior Month Data Collection

1. Calculate prior month dates - first day to last day of previous calendar month
2. Get accurate count - verify not hitting limits for the month
3. Extract complete dataset - all PRs merged in prior month with metadata

Phase 2: Initial Classification

1. Separate bot vs human PRs
2. Count human contributors from direct PRs
3. Identify bot PR candidates for analysis

Phase 3: Bot PR Analysis

For small datasets (< 50 bot PRs):

• Analyze all bot PRs directly

For large datasets (> 50 bot PRs):

• Sample analysis first (10-20 PRs) to validate approach
• Batch processing (50 at a time) with progress monitoring
• Full analysis once methodology confirmed

Phase 4: Human Author Attribution

1. Extract commit data for each bot PR
2. Identify human contributors from commit authors
3. Map bot PRs to their real human authors
4. Handle edge cases: multiple authors, partial automation

Phase 5: Final Analysis & Report Generation

1. Combine statistics - direct PRs + bot PR attributions
2. Rank contributors by total contributions
3. Identify patterns - automation usage, collaboration styles
4. Generate insights - development velocity, tooling adoption
5. Save results - create markdown report in .reports/ directory

Sample Code Patterns

Getting Prior Month PR Data

# Calculate prior month date range
PRIOR_MONTH_START=$(date -d "$(date +%Y-%m-01) -1 month" +%Y-%m-01)
PRIOR_MONTH_END=$(date -d "$(date +%Y-%m-01) -1 day" +%Y-%m-%d)

echo "Analyzing prior month: $PRIOR_MONTH_START to $PRIOR_MONTH_END"

# Always check count first
TOTAL_PRS=$(gh pr list --state merged --search "merged:>=$PRIOR_MONTH_START merged:<=$PRIOR_MONTH_END" --json number --limit 1000 | jq 'length')
echo "Total PRs in prior month: $TOTAL_PRS"

# If count equals limit, increase limit or use pagination
if [ $TOTAL_PRS -eq 1000 ]; then
    echo "Warning: May have hit limit, consider pagination"
fi

# Get full dataset for prior month
gh pr list --state merged --search "merged:>=$PRIOR_MONTH_START merged:<=$PRIOR_MONTH_END" --json number,title,author,mergedAt,body --limit 1500

Analyzing Bot PR Commits

# For each bot PR
for pr_num in $bot_pr_numbers; do
    echo "Analyzing PR #$pr_num"
    gh pr view $pr_num --json commits
    # Process commit authors...
done

Saving Results

# Create reports directory if it doesn't exist
mkdir -p .reports

# Generate report filename with prior month and year
REPORT_DATE=$(date -d "$(date +%Y-%m-01) -1 month" +%Y-%m)
REPORT_FILE=".reports/pr-analysis-${REPORT_DATE}.md"

# Save comprehensive results
cat > $REPORT_FILE <<EOF
# PR Analysis Report: $REPORT_DATE

## Summary
- **Analysis Date**: $(date +%Y-%m-%d)
- **Period**: $PRIOR_MONTH_START to $PRIOR_MONTH_END
- **Total PRs**: $TOTAL_PRS
- **Bot PRs**: $BOT_PR_COUNT
- **Human-contributed Bot PRs**: $HUMAN_BOT_PR_COUNT

## Top Contributors
[Generated contributor rankings with direct + bot PR counts]

## Analysis Details
[Detailed breakdown of findings]
EOF

echo "Report saved to: $REPORT_FILE"

Common Patterns Discovered

Bot Collaboration Types

1. Bot Initiator: Bot creates PR, human iterates (most common)
2. Human Initiator: Human work routed through bot workflow
3. Pure Automation: Genuinely bot-only work (typically 20-50% of bot PRs)

Email Address Patterns

• Human emails: Real domains, personal addresses
• Bot emails: [number]+[botname][bot]@users.noreply.github.com
• System emails: [e-mail], generic test accounts

Development Insights

• Heavy bot users often most productive contributors overall
• Bot workflows enable rapid iteration and refinement
• True automation focuses on maintenance, dependencies, formatting
• Human-bot collaboration shows sophisticated tooling integration

Error Handling

Common Issues

• Rate limiting: Use delays between API calls if needed
• Invalid PRs: Some PR numbers may not exist, handle gracefully
• Empty commit data: Some PRs may have no commit info
• Mixed authorship: PRs with both human and bot commits

Validation Steps

• Cross-reference total counts with manual verification
• Sample-check commit analysis results
• Verify human names aren't misclassified bot accounts
• Confirm attribution makes logical sense

Usage Examples

Basic Monthly Analysis

/pr-authorship-analysis "Analyze all PRs from last month"

Prior Month Bot Analysis

/pr-authorship-analysis "Find real authors behind bot PRs from the prior calendar month"

Monthly Team Report

/pr-authorship-analysis "Generate monthly contributor report for the completed month"

Specific Month Analysis

/pr-authorship-analysis "Show me March 2026 PR statistics with real authors"

Expected Outcomes

This skill will provide monthly reports with:

• Prior month contributor rankings including previously hidden human work behind bot PRs
• Monthly development velocity with accurate human attribution
• Team productivity insights showing automation adoption patterns for the month
• Bot vs human collaboration patterns from the completed calendar month
• True monthly metrics properly attributing bot-assisted contributions

The analysis typically reveals that 50-80% of "bot" PRs actually contain significant human contributions, providing accurate monthly team performance data and proper attribution to developers using advanced AI-assisted workflows.

Output and Storage

Report Location: Results are automatically saved to .reports/pr-analysis-YYYY-MM.md where YYYY-MM represents the analyzed month.

Example Report Path:

• March 2026 analysis → .reports/pr-analysis-2026-03.md
• February 2026 analysis → .reports/pr-analysis-2026-02.md

Report Structure:

# PR Analysis Report: 2026-03

## Executive Summary
- **Analysis Date**: 2026-04-01
- **Period**: 2026-03-01 to 2026-03-31
- **Total PRs**: 739
- **Bot PRs**: 141 (19.1%)
- **Human-contributed Bot PRs**: 70 (50% of bot PRs)

## Top Contributors
1. **Christiaan Arnoldus**: 142 PRs (99 direct + 43 bot collaborations)
2. **eshurakov**: 82 PRs (78 direct + 4 bot collaborations)
3. **jrf0110**: 71 PRs (all direct)

## Bot Analysis Details
- Pure automation: 71 PRs
- Human collaboration: 70 PRs
- Most active bot collaborator: Christiaan Arnoldus (43 PRs)

## Team Insights
- Team velocity: 23.6 PRs per day
- Bot adoption: 19.1% of PRs use bot workflows
- Human-AI collaboration: 50% of bot PRs involve human work

The reports directory (.reports/) is automatically created and can be:

• Committed to git for historical tracking
• Added to .gitignore to keep reports local
• Shared via PR for team visibility