Tracer-Cloud/opensre ? reverse-engineered prompt

I want you to build me a command line tool called opensre that lets anyone run their own AI SRE agents for handling production incidents.

When I type opensre it should drop me into an interactive terminal session where I can just describe an alert in plain English like my mongodb cluster is dropping connections and it streams a live investigation using my connected tools. It pulls in logs metrics traces runbooks from datadog grafana honeycomb or whatever I have set up does root cause analysis asks clarifying questions if needed and suggests fixes.

Add an onboard command that walks me through connecting my llm provider plus slack alerts aws accounts and other observability tools. Support a quick investigate command that takes a json alert file and runs the same flow non interactively.

Make the whole thing feel like a smart incident response copilot that stays grounded in real data from my infra. It should be easy to install with a one liner script work locally or deploy to railway with postgres and redis and include some basic synthetic test incidents so I can try it out right away. Use python and whatever modern agent frameworks make sense. Look up current best practices for tool calling if you need to.