GitHub Issue Duplicate Detection Skill

This skill provides patterns for detecting and handling duplicate GitHub issues.

Why Check for Duplicates

• •Prevent duplicate work
• •Consolidate discussion
• •Better tracking and metrics
• •Cleaner backlog

Detection Process

Step 1: Extract Key Terms

From the potential new issue:

• •Core functionality/feature
• •Error messages (exact text)
• •Component names
• •User actions described

Step 2: Search Existing Issues

# Search open issues
gh issue list --state open --search "keyword1 keyword2" --limit 20

# Search all issues (including closed)
gh issue list --state all --search "keyword1 keyword2" --limit 20

# Search with specific terms
gh issue list --state all --search "in:title error message"

# Search by label
gh issue list --state open --label "bug" --search "keyword"

Step 3: Review Results

For each potential match:

• •Compare titles carefully
• •Read descriptions
• •Check if closed issue was fixed
• •Consider different wordings of same problem

Step 4: Present Findings

## Duplicate Check Results

### Potential Duplicates Found

| # | Title | State | Created | Similarity |
|---|-------|-------|---------|------------|
| #123 | Similar issue title | Open | 2024-01-10 | High |
| #456 | Related problem | Closed | 2023-12-01 | Medium |

### Analysis

**#123** appears to describe the same issue:
- Same error message mentioned
- Same component affected
- Both triggered by same user action

**Recommendation**: Add context to #123 instead of creating new issue.

### If Not Duplicate

If you confirm this is a new issue, I'll proceed with creation.

Search Strategies

By Error Message

gh issue list --state all --search "TypeError: Cannot read property"

By Component/Feature

gh issue list --state all --search "in:title checkout"
gh issue list --state all --search "in:body payment"

By Author (Recent)

gh issue list --state all --author @me --search "keyword"

Combined Search

gh issue list --state open --label "bug" --search "in:title login error"

Similarity Assessment

High Similarity (Likely Duplicate)

• •Same error message
• •Same steps to reproduce
• •Same component affected
• •Different wording, same problem

Action: Recommend adding to existing issue

Medium Similarity (Possibly Related)

• •Same area but different specifics
• •Could share root cause
• •Related symptoms

Action: Recommend linking issues

Low Similarity (Probably Different)

• •Same component, different issue
• •Superficially similar only

Action: Proceed with new issue

Handling Duplicates

When Duplicate Found

1. •Don't create new issue
2. •Add context to existing:

gh issue comment 123 --body "$(cat <<'EOF'
Additional report:

[New information not in original issue]

Confirming this is still occurring as of [date].
EOF
)"

1. •If existing is closed:
   • •Check if fix actually resolved it
   • •May need to reopen with new info

Linking Related Issues

When related but not duplicate:

In new issue body:
Related to #123
See also #456

Closing as Duplicate

gh issue close {number} --comment "Closing as duplicate of #123"
gh issue edit {number} --add-label "duplicate"

Common Duplicate Patterns

Same Bug, Different Reports

• •Different users reporting
• •Slightly different reproduction steps
• •Same underlying cause

Feature Requests

• •Multiple requests for same feature
• •Different use cases, same solution
• •Different wording, same goal

Regression Reports

• •Previously fixed, returned
• •Same symptoms as old issue
• •May warrant new issue with reference

Search Query Reference

Combine with AND:

gh issue list --search "is:open label:bug in:title checkout"

Checklist Before Creating

• • Searched for exact error messages
• • Searched for key feature terms
• • Checked component-specific issues
• • Reviewed recent issues
• • Verified no duplicate exists
• • OR found duplicate and added context