Problem
As the server is currently running on a local system it is not available 24/7 and therefore not able to utilise usage-limits to its best. Also it does not live up to the promise to "work while you are asleep".
Solution
As a solution we should move all of this into a Dockerfile so we can build an image that can be run on server infrastructure and is only accessed using the AutoPR-WebUI.
The Dockerimage should provide the base for a multistage build that includes
- A LLM-Provider cli image, eg. claude-code-cli
- The Auto-PR image, based on alpine with a minimal set of available tools (like sh, curl)
- The "configured image" that takes both of the others adds the prompts, configuration and external tooling to it.
This can then be run with secrets being injected using environment variables.
Problem
As the server is currently running on a local system it is not available 24/7 and therefore not able to utilise usage-limits to its best. Also it does not live up to the promise to "work while you are asleep".
Solution
As a solution we should move all of this into a Dockerfile so we can build an image that can be run on server infrastructure and is only accessed using the AutoPR-WebUI.
The Dockerimage should provide the base for a multistage build that includes
This can then be run with secrets being injected using environment variables.