Files
metabuilder/scripts/README-triage.md

7.3 KiB

Duplicate Issue Triage Script

Overview

The triage-duplicate-issues.sh script is a smart automated tool that finds and closes duplicate issues in the repository. Unlike manual triage, this script:

  • Auto-detects all duplicate issue titles without manual configuration
  • Handles multiple groups of duplicates in a single run
  • Keeps the most recent issue open for each duplicate group
  • Adds explanatory comments before closing duplicates
  • Supports dry-run mode for safe testing

Problem It Solves

When automated systems create multiple issues with the same title (e.g., deployment failures), you end up with many duplicate issues that clutter the issue tracker. This script automatically detects and closes them, keeping only the most recent one.

Before

Issues:
  #199 ⚠️ Pre-Deployment Validation Failed (most recent)
  #195 ⚠️ Pre-Deployment Validation Failed (duplicate)
  #194 ⚠️ Pre-Deployment Validation Failed (duplicate)
  ... 26 more duplicates

After

Issues:
  #199 ⚠️ Pre-Deployment Validation Failed (open)
  #195 ⚠️ Pre-Deployment Validation Failed (closed - duplicate)
  #194 ⚠️ Pre-Deployment Validation Failed (closed - duplicate)
  ... 26 more closed with explanation

Usage

Basic Usage (Auto-detect all duplicates)

export GITHUB_TOKEN="ghp_your_token_here"
./scripts/triage-duplicate-issues.sh

This will:

  1. Fetch all open issues in the repository
  2. Group them by exact title match
  3. For each group with 2+ issues, close all except the most recent
  4. Add a comment explaining why each duplicate was closed

Dry Run (Preview without closing)

Always test with dry-run first!

export GITHUB_TOKEN="ghp_your_token_here"
./scripts/triage-duplicate-issues.sh --dry-run

This shows exactly what would be closed without actually closing anything.

Filter by Specific Title

If you only want to close duplicates of a specific title:

export GITHUB_TOKEN="ghp_your_token_here"
export SEARCH_TITLE="⚠️ Pre-Deployment Validation Failed"
./scripts/triage-duplicate-issues.sh

Get Help

./scripts/triage-duplicate-issues.sh --help

How It Works

1. Fetch All Open Issues

The script fetches all open issues using the GitHub API, handling pagination automatically.

2. Group by Title

Issues are grouped by exact title match. Only groups with 2+ issues are considered duplicates.

3. Sort by Date

Within each group, issues are sorted by creation date (newest first).

4. Close Duplicates

For each group:

  • Keep the most recent issue open (canonical issue)
  • Close all older duplicates
  • Add an explanatory comment with a link to the canonical issue

5. Summary Report

At the end, the script shows:

  • Number of duplicate groups processed
  • Total number of issues closed
  • Summary of what was done

Example Output

🤖 Smart Duplicate Issue Triage
===============================

🔍 Fetching all open issues from repository...
📊 Found 31 total open issues

🔎 Automatically detecting duplicate titles...
🎯 Found 1 title(s) with duplicates

🔧 Starting bulk issue triage...

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📋 Processing duplicate group 1/1
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Title: "⚠️ Pre-Deployment Validation Failed"

  📊 Found 29 issues with this title
  📌 Most recent: Issue #199 (created: 2025-12-27T18:12:06Z)

  🎯 Planning to close 28 duplicate issues

  📝 Adding comment to issue #195...
  ✅ Added comment to issue #195
  🔒 Closing issue #195...
  ✅ Closed issue #195

  [... continues for all duplicates ...]

  ✅ Completed processing this duplicate group

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
✨ Triage complete!
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📊 Summary:
  • Processed 1 duplicate title group(s)
  • Closed 28 duplicate issue(s)
  • Kept the most recent issue open for each title

Requirements

  • bash 4.0+
  • curl (for GitHub API calls)
  • jq (for JSON parsing)
  • GitHub Personal Access Token with repo scope

Testing

The script includes comprehensive tests:

./scripts/test-triage-logic.sh

This runs 8 test cases covering:

  • Smart duplicate detection
  • Multiple duplicate groups
  • Title filtering
  • Edge cases (single issue, empty input, no duplicates)

Safety Features

  1. Dry-run mode: Test before closing anything
  2. API error handling: Graceful failure on API errors
  3. Pagination: Handles repositories with 100+ issues
  4. Explanatory comments: Each closed issue gets a comment explaining why
  5. Rate limiting: 1-second delay between closures to avoid API limits
  6. Most recent preserved: Always keeps the newest issue open

Common Use Cases

Automated Deployment Failure Issues

When CI/CD creates multiple issues for deployment failures:

export GITHUB_TOKEN="ghp_xxxx"
export SEARCH_TITLE="🚨 Production Deployment Failed"
./scripts/triage-duplicate-issues.sh --dry-run  # Preview first
./scripts/triage-duplicate-issues.sh            # Then execute

Clean Up All Duplicates

If your repository has multiple types of duplicate issues:

export GITHUB_TOKEN="ghp_xxxx"
./scripts/triage-duplicate-issues.sh --dry-run  # Preview all
./scripts/triage-duplicate-issues.sh            # Close all

Scheduled Cleanup

Add to cron or GitHub Actions:

# .github/workflows/triage-duplicates.yml
name: Triage Duplicate Issues
on:
  schedule:
    - cron: '0 0 * * 0'  # Weekly on Sunday
  workflow_dispatch:       # Manual trigger

jobs:
  triage:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - name: Triage duplicates
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: ./scripts/triage-duplicate-issues.sh

Troubleshooting

"Bad credentials" error

Make sure your GITHUB_TOKEN has the repo scope and is not expired.

"jq: command not found"

Install jq:

# macOS
brew install jq

# Ubuntu/Debian
sudo apt-get install jq

# RHEL/CentOS
sudo yum install jq

No duplicates found

The script only detects issues with exact title matches. Similar but not identical titles won't be grouped together.

Rate limiting

If you hit rate limits, the script includes a 1-second delay between API calls. For large batches, you may need to wait or increase the delay.

Contributing

Improvements welcome! Some ideas:

  • Support fuzzy title matching (similar but not exact)
  • Add interactive mode to confirm each closure
  • Support closing by label or other criteria
  • Add GitHub Actions integration

License

Same as the repository (see root LICENSE file).