Your Repo's Issue Queue Is a Mess. Here's a Free Model Triage Workflow That Actually Runs

说明与用途

使用免费 AI 模型和免费服务器实现 GitHub issue 自动分类，按四类标签给出评论建议。

问题

Maintainers struggle with large, unstructured issue queues, spending time manually sorting issues into categories such as bug, feature, question, or duplicate.

结果

In a test repo with ~40 open issues, the model correctly labeled 34 issues; the remaining 6 were ambiguous. The workflow runs entirely on free tiers, requiring no GPU or credit card.

工作流

• 1. Configure a free server (e.g., MonkeyCode) to host the script.
• 2. Set environment variables MONKEYCODE_API_URL and MONKEYCODE_API_KEY.
• 3. Deploy the Python script that receives GitHub webhook payloads.
• 4. Register a GitHub webhook for the repository with payload URL pointing to the server, content type application/json, and event Issues.
• 5. When a new issue opens, GitHub sends payload to the server.
• 6. The handler extracts title and body, truncates body to 2000 chars, and builds a prompt.
• 7. The script calls the free model with temperature 0 and max_tokens 10, requesting one of four labels.
• 8. The model returns a single word (bug, feature, question, duplicate).
• 9. The script posts a comment on the issue with the suggested label.
• 10. Maintainer reviews the comment and applies the label if desired.

证据

I tested this workflow against a small repo with about 40 open issues. The model correctly labeled 34 of them.

来源

The goal is small. A webhook receives a new issue. The script pulls the issue title and body. It sends both to a free model and asks for one of four labels: bug, feature, question, or duplicate.

来源