create-issue

Convert failing GitHub Actions runs into deduplicated NVIDIA/Megatron-LM issues.

17.4k|4.3k|Updated Mar 21, 2019

Search Tags:#github #github actions #log analysis #bug reporting #issue creation #ci triage #pytest failures

Ready to run

npx skills add https://github.com/NVIDIA/Megatron-LM --skill create-issue-nvidia

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: create-issue
Source: https://github.com/NVIDIA/Megatron-LM/tree/main/skills/create-issue
Command: npx skills add https://github.com/NVIDIA/Megatron-LM --skill create-issue-nvidia

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you quickly diagnose a failing GitHub Actions run and turn the failure evidence into a clear, reproducible bug report that maintainers can act on.

Core Features & Use Cases

Parse Actions URLs and select failures: Accepts a run or job URL, identifies failed jobs, and supports triaging one job or multiple failed jobs.
Extract actionable root-cause evidence: Retrieves job logs, isolates error-bearing lines, captures failed test node IDs, and summarizes the most relevant traceback content without guessing.
Ensure deduplication and correct attribution: Searches for existing open issues to avoid duplicates and assigns the new issue to the most recent author of the failing test file when possible.
Use for: When you share a GitHub Actions URL after a CI regression, and you want the issue filed against NVIDIA/Megatron-LM with the right links, labels, and minimal error snippets.

Quick Start

Provide the failing GitHub Actions run or job URL (job URL preferred) and ask the assistant to create an issue for that specific CI failure.

Frequently Asked Questions about create-issue

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I create a GitHub issue from a failing GitHub Actions run?▼

To create a GitHub issue from a failing GitHub Actions run, provide the run or job URL to automatically extract logs, isolate failed pytest node IDs, and file a deduplicated bug report with proper labels.

Can I automatically extract failed pytest node IDs from CI logs to report bugs?▼

Yes, you can automatically extract failed pytest node IDs from CI logs by parsing the provided GitHub Actions job URL, filtering the retrieved logs for error-bearing lines, and summarizing the relevant traceback content.

Does this issue creation process check for duplicate GitHub issues before filing?▼

Yes, the issue creation process checks for duplicate GitHub issues by searching existing open issues before creating a new one, ensuring proper deduplication and linking the appropriate issue with correct attribution.

How do I attribute a CI regression bug to the right test author in GitHub?▼

To attribute a CI regression bug to the right test author in GitHub, the system resolves the triggering PR and identifies the most recent author of the failing test file via GitHub APIs before assigning the issue.

What's the best way to triage multiple failed jobs in a GitHub Actions run?▼

The best way to triage multiple failed jobs in a GitHub Actions run is to supply the run URL, which parses the run_id to identify all failed jobs and supports triaging one job or multiple failed jobs simultaneously.

Are there limitations when converting GitHub Actions job URLs into bug reports?▼

A limitation when converting GitHub Actions job URLs into bug reports is that the parsing and attribution specifically target NVIDIA/Megatron-LM repositories, requiring valid run_id or job_id extraction to fetch and filter job logs accurately.