chat.tux
Terminal UX (TUX) for the Agent Runtimes Chat assistant.
STYLE_PRIMARY
Green accent 1ABC9C - primary accent in dark mode
STYLE_SECONDARY
Green brand 16A085 - secondary accent
STYLE_ACCENT
Green bright 2ECC71 - highlights
STYLE_MUTED
Gray 59595C - supporting text
STYLE_WHITE
Primary text in dark mode
STYLE_ERROR
Error states
STYLE_WARNING
Warning states
SlashCommandCompleter Objects
class SlashCommandCompleter(Completer)
Completer for slash commands with menu-style display.
get_completions
def get_completions(document: Any, complete_event: Any) -> Any
Yield completions for slash commands.
ToolCallInfo Objects
@dataclass
class ToolCallInfo()
Information about a tool call.
status
in_progress, complete, error
format_args
def format_args(max_value_len: int = 40) -> str
Format arguments for display.
SessionStats Objects
@dataclass
class SessionStats()
Session statistics for token tracking.
CliTux Objects
class CliTux()
Terminal UX for the Agent Runtimes Chat assistant.
__init__
def __init__(agent_url: str,
server_url: str = "http://127.0.0.1:8000",
agent_id: str = "chat",
eggs: bool = False,
jupyter_url: Optional[str] = None,
extra_suggestions: Optional[list[str]] = None)
Initialize the TUX.
Arguments:
agent_url- URL of the AG-UI agent endpointserver_url- Base URL of the agent-runtimes serveragent_id- Agent ID for API callseggs- Enable Easter egg commandsjupyter_url- Jupyter server URL (only set when sandbox is jupyter)extra_suggestions- Additional suggestions provided via --suggestions flag
show_welcome
def show_welcome() -> None
Display the welcome banner.
show_prompt
async def show_prompt() -> str
Display the prompt and get user input with slash command completion.
handle_command
async def handle_command(user_input: str) -> Optional[str]
Handle a slash command.
Returns:
None if no command matched or the command produced no follow-up. A non-empty string when a command returns a prompt to send to the agent (e.g. /suggestions returns the chosen suggestion text). The empty string "" signals the command was handled but has no follow-up.
send_message
async def send_message(message: str) -> None
Send a message to the agent and stream the response.
run
async def run() -> None
Run the main TUX loop.
run_tux
async def run_tux(agent_url: str,
server_url: str = "http://127.0.0.1:8000",
agent_id: str = "chat",
eggs: bool = False,
jupyter_url: Optional[str] = None,
extra_suggestions: Optional[list[str]] = None) -> None
Run the Agent Runtimes Chat assistant TUX.
Arguments:
agent_url- URL of the AG-UI agent endpointserver_url- Base URL of the agent-runtimes serveragent_id- Agent ID for API callseggs- Enable Easter egg commandsjupyter_url- Jupyter server URL (only set when sandbox is jupyter)extra_suggestions- Additional suggestions provided via --suggestions flag