`class Shell`

A pseudo-LLM provider that executes user input as code via the existing Brute::Tools::Shell tool.

Models correspond to interpreters:

bash - pass-through (default) ruby - ruby -e '...' python - python3 -c '...' nix - nix eval --expr '...'

The provider's #complete method returns a synthetic response containing a single "shell" tool call. The agent loop executes it through the normal pipeline — all middleware (message tracking, session persistence, token tracking, etc.) fires as usual.

Nested

ModelList

Definitions

`def models`

For the REPL model picker: provider.models.all.select(&:chat?)

Implementation

def models
  ModelList.new(MODELS)
end

`def extract_text(prompt)`

Extract the user's text from the messages array. Returns nil when the messages contain tool results (the second round-trip) so #complete knows to return an empty response.

Implementation

def extract_text(prompt)
  case prompt
  when String
    prompt
  when ::Array
    return nil if prompt.any? { |m| m.respond_to?(:role) && m.role == :tool }

    user_msg = prompt.reverse_each.find { |m| m.respond_to?(:role) && m.role == :user }
    user_msg&.content.to_s
  else
    prompt.to_s
  end
end