Drops you into an interactive chat session against the ExAthena agent loop.
Defaults to the provider set in config :ex_athena, default_provider: :ollama
(falls back to :ollama when unset). The model is read from the provider's
config block, e.g. config :ex_athena, :llamacpp, model: "my-model".
For a full walkthrough — keybindings, mode switching, JSON provider setup, and screenshot tour — see the TUI guide.
Usage
mix athena.chat
mix athena.chat --provider llamacpp
mix athena.chat --provider ollama --model qwen2.5-coder:14b
mix athena.chat --mode plan_and_solve
mix athena.chat -p ~/projects/my_appFlags
--provider NAME—:ollama,:llamacpp,:openai, etc. (overrides config).--model NAME— initial model (overrides config).--mode NAME—react,plan_and_solve, orreflexion.--path PATH(-p) — working directory for tools that touch the filesystem.~is expanded. The chat process does NOTcdinto this path; tools just receive it as theircwd.
Provider-specific requirements
:ollama— requires a running Ollama daemon (ollama serve).:llamacpp— requires a running llama.cpp server (llama-server --model ...).