I have been toying around with the idea of using agents for running chaos engineering experiments.

The idea of this project is as follow:

- There prebuild skills for different target systems. The idea is to test and maintain these skills - You bring your own LLM - You just specify the experiment in natural language, the agent determines the skills to be used, plans and executes the experiments, rolls back if any configuration changes were made.

An experiment might looks as follows:

``` chaos run "run resilience test on the database postgres://postgres:mysecretpassword@localhost:5432/postgres for 10s" ```