Merge Duplicate Issues Without Losing Context

Where this fits

You're a maintainer juggling two open issues that describe the same bug from different angles. One says "crash on startup", the other says "fails to launch with config X". You know they're the same root cause, but you don't want to lose either reporter's details. Manually closing one, copying notes, and cross-referencing is tedious. This is where magpie-issue-deduplicate steps in—it automates the proposal, not the decision.

Why agents benefit

• Preserves both reporters' context: The skill generates a closing comment that summarizes the duplicate's unique details and a cross-reference comment on the kept issue, so no information is lost.
• Waits for explicit confirmation: It never posts or closes anything without a maintainer's go-ahead, keeping the human in control of irreversible actions.
• Understands natural language triggers: You can say "deduplicate #123 and #456" or "#456 is the same as #123"—the agent parses intent and proposes the merge.
• Integrates with triage workflows: It pairs with issue-triage or issue-stale-sweep to catch likely duplicates before they pile up.
• Avoids false merges: It knows when to skip (e.g., security reports, or issues that only share a surface) and suggests cross-linking instead.

Practical scenarios

Scenario 1: Maintainer spots a duplicate pair

A maintainer sees two open issues about a memory leak in the same module. They tell the agent: "close #210 as a duplicate of #198". The skill drafts a closing comment for #210 that includes the unique stack trace from that reporter, and a comment on #198 linking back. The maintainer reviews and approves—then the agent posts both.

Scenario 2: Triage bot surfaces a likely match

Your issue-stale-sweep agent flags #340 and #341 as possible duplicates because they share keywords and similar symptoms. It invokes magpie-issue-deduplicate with the kept issue and the duplicate. The skill produces a proposal with a summary of overlapping evidence, ready for human review.

Scenario 3: Two reporters describe the same feature request

Two users request the same API endpoint but with different naming conventions. The skill recognizes the underlying need is identical, proposes merging them, and ensures both users' use cases are captured in the kept issue's cross-reference comment.

Add it to your agent workflow

Install the skill with:

npx skills add apache/magpie --skill magpie-issue-deduplicate

Then invoke it in your agent prompt or orchestration logic. Example usage:

User: "deduplicate #45 and #67"
Agent: I'll propose merging #67 into #45. I'll draft a closing comment for #67 and a cross-reference on #45. Awaiting your confirmation.

The agent will present the proposed comments and wait for your explicit approval before posting anything.

Compare before adopting

Before integrating, evaluate the skill's quality signals: check the GitHub repo (87 stars, Apache license) for recent commits and issue responsiveness. Compare it with alternatives like manual deduplication scripts or other agent skills that might auto-close without confirmation. Consider whether your workflow requires the human-in-the-loop guarantee—this skill's core value is that it never acts unilaterally.

Why it is worth tracking

magpie-issue-deduplicate is a practical, safety-first automation for a common maintainer pain point. Its Apache stewardship and clear when_to_use guidance indicate a mature design. As agent-driven triage becomes more common, skills that respect human oversight will gain traction. Keep an eye on its adoption and community feedback—if you handle many duplicate issues, this could become a staple in your agent toolkit. For more details, see the skill page.