How I Cleared a 214-Issue Backlog on My Open-Source Repo With Claude Code

Описание и применение

В течение трёх недель с помощью Claude Code я очистил 214‑й backlog открытого репозитория, классифицировав, проверив воспроизводимость и исправив 19 реальных багов, оставив только 9 актуальных вопросов.

Задача

A mid-sized open-source CLI library had 214 open issues, many unlabeled, duplicates, and stale, making maintenance difficult and risking loss of real bugs.

Результат

Three weeks of AI-assisted triage reduced the backlog from 214 to 9 issues, uncovered 19 live bugs, and improved community perception without mass-closing.

Рабочий процесс

• 1. Dump all open issues to a JSON file using gh CLI.
• 2. Prompt Claude Code to classify each issue into category, status_guess, confidence, and evidence.
• 3. Verify likely-fixed issues by running reproduction steps against current main and recording output.
• 4. Draft replies for all issues with a rule that the agent never posts; review drafts in batches and post approved comments.
• 5. Fix confirmed bugs using reproduction tests as regression tests, reviewing diffs before merging.

Доказательства

The first pass took the agent a couple of hours of chugging through the codebase and git history. Results on the 214 issues: ...

Источник

For issue #N: extract the reproduction steps from the report. Run them against current main in a scratch directory. Record: REPRODUCED / NOT-REPRODUCED / CANNOT-ATTEMPT (missing info). Save the actual terminal output as proof.

Источник