Video and materials released: GitHub Webinar Summer Vacation Special held on August 20th

In this webinar, we summarized the updates announced in the Changelog by product, such as GitHub Copilot, Actions, and Security. In addition, we welcomed JFrog as a guest and introduced the integration between GitHub and JFrog and the JFrog Copilot Extension.

Download presentation slides below:
PDF:WebinarSlidesAug20_GitHubWebinarSummerSpecial
JP_GitHubAndJFrogIntegration

Q&A

Q. Starting this Monday, when I assign Copilot to an issue, an internal error occurs almost 100% of the time. The error doesn't occur immediately, but usually occurs within about 3 minutes. Is this an issue that only occurs in our environment?
A. We apologize for the inconvenience, but please contact support so that we can make a decision based on the details of your situation. https://support.github.com/

Q. Are there any plans to support AGENTS.md?
A. We currently support the GitHub Copilot coding agent (falling back to AGENTS.md if copilot-instructions.md is missing), and we're considering supporting other areas as well.

Q. Regarding assigning tasks to agents using issues, I think that instead of using the MCP server, it will consume GitHub Actions. Is it possible to achieve the same results using GitHub and Playaround's MCP server in a VSCode environment without using GitHub Actions?
A. As you may be aware, if you use the coding agent, you will also be charged for GitHub Actions. If you don't use Actions, you can use GitHub Copilot's agent mode in VS Code, etc. Since you can use the MCP server in agent mode, you can achieve the same results by setting up Playwright MCP.

Q. Is it correct to understand that you decided that Playwright's MCP is okay for GitHub? I would appreciate it if you could tell me why you decided to use Playwright.
A. Playwright and Playwright MCP are developed by Microsoft, so you can use them with confidence.

Q. Please provide me with the link again where I can find the details of the Agreement to which the Pre-Release Terms apply.
A. We apologize for the English translation, but the updates to the pre-release terms and conditions, as well as who applies to them, are detailed here: https://github.blog/changelog/2025-08-08-defense-of-third-party-claims-added-for-volume-licensing-customers/

Q. I am working on a business that supports the introduction of Github Copilot. I would appreciate it if you could tell me the requirements for becoming a Github Copilot ambassador.
A. GitHub does not officially operate a program called GitHub Copilot Ambassador. GitHub Stars is a similar program, so please see the website for more information. https://stars.github.com/

Q. I've used up all my premium requests for this month, but even when I select GPT-5 mini or GPT-4o, which are both 0x, I'm still falling back to GPT 4.1. Is this by design?
A. We apologize for the inconvenience, but we have identified this as a bug and will be fixing it in the future.

Q. I have a question about copyright. If the blocking feature is set to ON, I understand that GitHub will provide legal compensation if a code suggestion generated by GitHub Copilot is sued for copyright infringement. However, if the blocking feature is turned OFF, is it correct that GitHub will not provide legal compensation if a copyright infringement lawsuit is filed?
A. The terms and conditions are listed below, so please refer to them: https://learn.microsoft.com/ja-jp/azure/ai-foundry/responsible-ai/openai/customer-copyright-commitment

Q. How does billing for codespaces work? Is it separate from the Copilot billing plan? I'm asking because it's possible to create a codespace via Spark -> Open Codespace.
A. Codespaces' pricing structure is determined by two factors: how long you use your Codespace environment and how much storage you consume. The usage fee is charged based on the resources of your Codespace environment. Storage fees apply regardless of whether the Codespace environment is running or not, unless you delete it. For more information, see the following:
https://docs.github.com/ja/billing/concepts/product-billing/github-codespaces

Q. Are there any plans for GitHub Copilot to support spec-driven development like AWS Kiro? (I previously saw a mention of Task lists in agent mode in the GitHub Changelog, so I thought that might be spec-driven.)
A. As you may know, we're working on an experimental feature in Copilot for VS Code that lets you create task lists and progress as you work.
https://code.visualstudio.com/updates/v1_103#_track-progress-with-task-lists-experimental
Additionally, we're also introducing a custom chat mode, which allows you to create plans and architectures and then move on to implementation.
https://github.com/github/awesome-copilot?tab=readme-ov-file#-custom-chat-modes

Q. Can github spark be integrated with middleware (DB) etc.?
A. Unfortunately, GitHub Spark does not support integration with middleware, etc.

Q. Does the Copilot agent need a Codespace subscription to use the execution environment? Can I use a local execution environment that is not a Codespace?
A. Using Codespaces is not required to use the GitHub Copilot coding agent, but it does require the use of GitHub Actions, as it runs on the GitHub Actions runner. However, there is also an agent mode available from Visual Studio Code, etc., which runs the agent in a local environment, so you can use the local execution environment.

Q. I would like the Enterprise Owner to obtain and manage the premium request consumption status of each Enterprise Member via API, rather than via the usage report csv, but I am not aware of any such API at present (consumption can be seen on an organization or repository basis). Are there any plans to release an API in the future that will allow me to check premium request consumption on a per-user basis?
A. There are no clear plans at this time.

Q. When using VSCode, I often receive command suggestions from Copilot, execute them in the VSCode terminal, and have Copilot read the response and suggest the next step. However, in a WSL environment, the command response often cannot be read correctly. Is VSCode + WSL + Github Copilot not supported?
A. Based on the information you've provided, it appears this issue may be related. We would appreciate your patience as we wait for an update.
https://github.com/microsoft/vscode/issues/252524

Q. I updated VS Code to the latest version, but I still can't use GPT-5-mini. Is there an environment dependency?
A. This depends on the Copilot plan you are using, but if you are using the Business or Enterprise plan, your administrator will need to opt in to use it. For more information, please see below:
https://github.blog/changelog/2025-08-13-gpt-5-mini-now-available-in-github-copilot-in-public-preview/#enabling-access

Q. I apologize if this is a bit off topic. I've been hearing a lot about something called github cpilot Beast Mode lately. What can Beast Mode do, and what are its benefits? Also, I'd be grateful if you could tell me any recommended ways to catch up.
A. Beast Mode is an example of a custom chat mode feature in VS Code's Copilot Chat. Custom chat modes allow you to customize Copilot's behavior. For example, you could create a planning mode that allows Copilot to only read operations.
https://code.visualstudio.com/docs/copilot/chat/chat-modes
Beast Mode is a custom chat mode designed to encourage proactive trial and error without abandoning tasks. Other examples of custom chat modes are available here, so please take a look.
https://github.com/github/awesome-copilot?tab=readme-ov-file#-custom-chat-modes

Q. I've identified four ways to run the Coding Agent.Which one do you recommend?

・The provided method via a github.com issue.
・Triggering from the issue screen of the vscode github pull request extension.
・Triggering from the vscode agent mode prompt using the delegate coding agent function.
・Triggering from the agent panel next to the copilot icon on github.com.
A. There are pros and cons to each. Creating an issue has the advantage that, for tasks that involve a certain degree of uncertainty or are larger in scale, you can describe in detail what you want to do and your policy in the issue, discuss it on the issue, and then assign the task to a coding agent. Also, if you're using issues as a means of receiving tasks or bug reports from outside the team, you can also use that flow to assign some tasks to a coding agent.
On the other hand, when requesting tasks from a coding agent via Copilot Chat or the agent panel on github.com, you'll likely be more likely to be casually requesting smaller tasks or tasks with a high degree of certainty.

Q. I'm considering using SBOM (Software BOM) to manage source code versions, dependencies, etc. Can a central repository do this?
A. (Answer from JFrog) Yes, we do.

Q. Our company is considering using GitHub Copilot to develop tools for our operations and mathematical departments. Could you please tell me if there are any versions that are commonly adopted for use by people other than the systems department?
A. GitHub Copilot is available in two main types: individual and organizational. These include:

Copilot Pro and Pro+
Individual contracts.
These are often adopted for small-scale use or for trial purposes.
Copilot Business and Enterprise
Organization-based contracts. These are widely used by departments other than systems (such as operations and mathematical departments), and many companies choose them for their user management, security, and billing management.
In particular, for corporate use, Copilot Business is almost always chosen, regardless of department.

Q. What exactly does the Copilot agent do within the runtime environment?
A. Cloning a repository into an execution environment allows the agent to access the code, and when the agent builds or tests the code, the process is carried out in this execution environment.