llm-books/sre-agent ? reverse-engineered prompt
Reverse engineered prompt
Build me a local demo of an AI SRE agent that can watch a small fake ecommerce system, notice when something is broken, investigate it with logs, metrics, and traces, then suggest what went wrong and how to fix it. I want the whole thing to run on Docker Desktop on a laptop, with six services, a checkout style flow, a load generator, a chaos tool that can deliberately break things, and a dashboard setup so I can see incidents happen in real time.
The agent should keep a durable record of what it has already investigated, so if I run it again on the same incident it can replay the prior result instead of starting over. It should be able to work fully offline with a scripted planner, but also support a real model if I give it an API key. Make it feel like a book companion project I can run chapter by chapter, with simple make commands to start the environment, inject a failure, and let the agent handle it. If you need to check current docs for any tools, go ahead and look them up online.
Are you gonna build this?
make sure you review the code using arcumet