[FEATURE]: Duplicate Issue Detection and labeler as `Possible Duplicate` using semantic analysis

[aniket866]

Feature and its Use Cases

I explored two possible approaches for implementing a duplicate issue detector that labels new issues as “possible duplicate” with an X% similarity score.

This issue is to compare both approaches before finalizing implementation at the org level.

Problem Statement

• **Currently, contributors must manually search through existing issues before opening a new one.

• This is time-consuming and still leads to duplicate issues.**

• We need an automated way to:

1. Detect similar issues when a new issue is opened

2. Suggest possible duplicates (without auto-closing)

3. Label them clearly.

4. The system should assist, not enforce.

### Approach 1: Using CodeRabbit (Issue Enrichment)

• Org template: Template-Repo/.coderabbit.yaml

How it Works

• Configure duplicate detection via .coderabbit.yaml
• CodeRabbit enriches issues automatically
• Can be rolled out across org via template repo

Pros:

• Centralized configuration (via template repo)

• Org-wide scalability

• No need to maintain custom workflows per repo

• AI-powered enrichment (context-aware analysis)

• Cleaner long-term maintenance

Cons

• May not allow fine-grained control over scoring logic

• Requires org-level alignment before rollout

Approach 2: Using GitHub Action Bot (Custom Workflow)

• Implemented example in PictoPy testing issue.

How it Works

• Custom GitHub Action triggers on issues: opened

• Compares issue title + body with existing issues

• Calculates similarity score

• Comments with similar issues

• Labels as possible duplicate

Pros

• Full control over logic and scoring

• Customizable similarity threshold

• Can mark as “X% duplicate”

• No dependency on third-party enrichment

• Transparent workflow

Cons

• Needs per-repo workflow unless added to template repo

• Maintenance burden on org

• AI capability depends on implementation quality

• Might require API rate handling

• Less intelligent compared to AI-native enrichment (unless enhanced)

Additional Context

this is for duplicate issue detector and labelling as possible duplicate ,using github action bot

• How: Use a GitHub bot and a simple GitHub Action to automatically check new issues and show similar existing ones.

• Impact: Duplicate issues get clearly marked, so maintainers don’t have to spend extra time checking or replying again.

• Notes: The system will only suggest duplicates and will not auto-close any issue.**

• Note : it marks as x% duplicate

• for more details checkout this testing issue BUG:A Time-of-Check to Time-of-Use (TOCTOU) race condition exists in sync-microservice/app/utils/watcher.py aniket866/PictoPy#5

so

Code of Conduct

• I have joined the Discord server and will post updates there
• I have searched existing issues to avoid duplicates