How to automatically chat with your co-pilot?

[weite76]

Select Topic Area

Question

Copilot Feature Area

Copilot in GitHub

Body

I frequently chat with my co-pilot and have it analyze program logs. I'd like to automate this chat, perhaps sending it hourly messages and having it write analysis reports to my project. I don't want to use the API because that requires me to handle the context myself.

[AvishekDas5]

You can’t fully automate chats here without the API, scheduled messaging and file output aren’t supported in the chat interface. For automation (hourly logs, reports to your project), you’ll need the API or a wrapper around it, since that’s the only way to programmatically maintain context and trigger actions.

Set up a cron (or Task Scheduler) job that runs a Python script hourly:

Read logs

Send them to GPT via the Chat Completions API

Save the response as a report

You can also feed past summaries back into the prompt to keep context.

[zephinax]

Hi @weite76,

Currently, GitHub Copilot (including Copilot Chat in VS Code) does not support fully automated messaging or scheduled chats. Copilot is designed to provide real-time suggestions and responses interactively, so sending it hourly messages automatically is not supported out-of-the-box.

A few points to consider:

Copilot Chat requires manual interaction: There’s no built-in scheduler.

API alternative: While you mentioned you want to avoid the API, it’s currently the only way to programmatically manage context and automate analysis. Without the API, the system can’t maintain conversation history or generate reports automatically.

Workarounds:

You could write scripts that prepare logs or analysis prompts and then use the API to submit them to a Copilot-like model.

For local automation, consider using GitHub Actions to periodically gather logs and generate files, then manually interact with Copilot for analysis.

Full chat automation isn’t supported yet. If your goal is automated log analysis, combining GitHub Actions + Copilot API (or another LLM API) is currently the practical approach.

[weite76]

Thanks, I didn't subscribe to any API.

[shondsouza]

Right now, Copilot Chat doesn’t support automatic scheduled conversations. It’s mainly designed for interactive use, so you’ll still need to trigger it manually.

That said, here are a few options you could try:

VS Code Agent mode: You can set up a custom command (like /analyze-logs) so Copilot can process your logs and generate reports with a single trigger.

GitHub Agents Panel (Mission Control): Lets you hand off tasks to Copilot directly on GitHub.com, but it’s still manual.

API: The only way to truly automate things (like hourly reports). The trade-off is you’ll need to handle the context yourself.

So for now there’s no “set it and forget it” option, but you can get close with Agent mode, or go fully automated with the API if you’re okay managing context.

[weite76]

It would be nice if I could set it to do something on a scheduled basis in the chat window. For example, I could set it to: Please reply to my instructions in the input box every hour. This way, the entire automated process can be realized in the chat window.

[weite76]

Because I often need it to do some monitoring and analysis work, I have to enter instructions over and over again. I hope this work can be automated.

[shondsouza]

Yeah, that would be super useful. Right now though, Copilot Chat can’t run on a schedule—it only works when you interact with it. The closest workaround is using Agent mode in VS Code (so you just trigger one command instead of retyping everything) or the API if you want true automation. Hopefully scheduled tasks get added in the future—it’d be perfect for your monitoring case.

[weite76]

Thank you! I hope it can become more like a human in the future, not just passively answering questions, but also proactively taking actions.