CodexNexor/klyra — reverse-engineered prompt

I want a self hosted AI security lab called Klyra that I can run on an Ubuntu or Debian box with one simple start command. The idea is that each user gets a fresh isolated workspace in LXD, and the AI testing agent runs inside that container instead of on the host. Give me a clean browser app with login, a chat based workspace, file and session history, command output, and tidy findings I can review later.

I also want an admin area where the owner can create users, promote access, inspect running workspaces, and stop or delete containers. Please keep the product clearly scoped to authorized testing only, with guardrails and wording that make it obvious this is for owned labs, internal appsec, CTFs, training, and written permission engagements.

Make the setup feel polished and practical, including sensible defaults, encrypted sensitive data, and an easy demo deploy that can optionally expose a temporary public URL. If you need details, check the current docs online and wire up whatever is needed to make the full app work end to end.