Kenta Kase

Japan

I'm an OSS developer focused on GitHub Actions and GitHub Copilot. I follow the GitHub blog's changelog daily, sharing the latest features to help others adopt new technologies.

Community Contributions

LatestBlog postsOpen sourceSpeaking

Trying Out GitHub’s New Stacked PRs: A Practical Guide to `gh stack`

I tested GitHub's new public preview feature, Stacked Pull Requests, by building a simple TODO application and breaking the development process into smaller, dependent PRs. Using the `gh stack` CLI extension, I explored essential commands like `gh stack submit`, `gh stack view`, and `gh stack modify` to manage branch layers and submit pull requests interactively. Overall, I found that `gh stack` provides an intuitive terminal-based interface that significantly simplifies managing and updating chains of stacked PRs.

Blogpost / 08-07-2026

actions-timeline v3.1.0: Visualizing Step-by-Step Breakdown for Composite Actions

I added a long-requested feature to my OSS tool, actions-timeline, allowing it to expand composite actions and display individual steps on the execution timeline. By taking inspiration from job log parsing techniques and utilizing AI coding assistants to automate testing loops, I successfully implemented this complex functionality. This update makes it significantly easier to identify exact performance bottlenecks within reusable GitHub Actions workflows.

Open source project / 04-21-2026

How to Semi-Automate Code Reviews with GitHub Copilot

To optimize my AI-driven development workflow while avoiding unnecessary automated reviews, I set up a semi-automated process to request GitHub Copilot code reviews conditionally. Instead of relying on GitHub Rulesets, I instruct my AI agent to trigger Copilot reviews using CLI commands like `gh pr edit` or GitHub MCP tools after creating a pull request. This approach eliminates unnecessary waiting time and allows me to receive reliable initial AI feedback before conducting my own manual review.

Blogpost / 02-16-2026

Syncing AI Agent Instructions: How to Make GitHub Copilot and Claude Code Work Together

I shared a practical method to synchronize custom instruction files and slash commands across multiple AI coding assistants, such as GitHub Copilot and Claude Code. By leveraging symbolic links, I managed to centralize various tool-specific configuration files into a single master source, ensuring consistent behavior regardless of the tool I use. This setup allows me to switch between different AI agents seamlessly while maintaining the same project-specific context and custom rules.

Blogpost / 12-19-2025

Beyond Coding: Using Issues and PRs for Systematic Solo Development with GitHub Copilot

I recently shared my detailed process for carrying out solo development by utilizing GitHub Copilot as a virtual teammate at VS Code Dev Days Tokyo. In this article, I walk through the full script of my presentation, illustrating how I collaboratively brainstorm feature specifications, systematically break down tasks into Issues and Sub-issues, and handle the implementation and code review stages. This methodical, Issue/Pull-request-driven approach enables me to manage my personal projects in a planned and structured way, just like engaging in a "standard" team development environment.

Blogpost / 09-18-2025

Stop Typing: Let GitHub Copilot Turn Meeting Notes into Perfect Issues

This article introduces a highly effective method I discovered for using GitHub Copilot Chat to create issues directly on GitHub. I explain that instead of manually typing out prompts, feeding the tool automatically generated meeting minutes—which typically contain "Next Action" items—significantly boosts my efficiency. This process enables Copilot to easily break down tasks, adhere to issue templates, and automatically set parameters like assignees, labels, and milestones, streamlining my task management workflow.

Blogpost / 07-12-2025

GitHub Copilot Utilization: A Guide for New Cybozu Developers

I recently learned how to leverage GitHub Copilot effectively in a professional setting, focusing on both its various features and the critical security considerations for business use. Specifically, I explored the functionalities available on github.com and within VSCode, including intelligent tab completion, Chat, and the powerful new Agent Mode. I was fascinated to see a demonstration of Agent Mode performing implementation, error correction, and testing almost entirely autonomously, which highlights its potential to significantly boost my development productivity.

Speaking (conference/usergroups) / 07-06-2025

What Can We Do with GitHub Copilot Spaces? A Review

I recently explored the newly released GitHub Copilot Spaces, which I feel is best described as GitHub's version of NotebookLM, allowing users to attach specific files and repositories as context for Copilot Chat. I found that its primary unique advantages over the standard Copilot Chat are currently limited to setting a system prompt and sharing the space within an organization, due to a strict token limit and the fact that the regular chat now supports multiple repository attachments. Despite these constraints, I believe this feature has potential for practical use cases, such as aggregating documentation for shared libraries or serving as a lightweight, alternative RAG system.

Blogpost / 05-30-2025

Related Stars

Kazumi OHIRA

dzeyelid

TAKASHI KAWAMOTO

mochan-tk

Tomoki Nomura

tmknom

Maki Nagase

yuma-722