Skip to main content

terminal_agent.main

What it is​

A terminal chat runner for a naas_abi_core.services.agent.Agent that:

  • Displays AI and tool output with Rich styling.
  • Accepts interactive input with a disappearing placeholder (TTY only).
  • Logs the terminal conversation to a timestamped file under storage/datastore/interfaces/terminal_agent/.

Public API​

  • init_conversation_file() -> pathlib.Path
    • Creates a new timestamped conversation log file and writes a header.
  • save_to_conversation(line: str) -> None
    • Appends a line to the current conversation log file (if initialized).
  • get_input_with_placeholder(prompt: str = "> ", placeholder: str = "Send a message (/? for help)") -> str
    • Reads user input.
    • Uses raw-terminal key handling for interactive sessions; falls back to input() for piped/non-TTY stdin.
  • on_tool_response(message: langchain_core.messages.AnyMessage) -> None
    • Formats and prints tool responses via print_tool_response.
    • Attempts to detect and render image paths (.png/.jpg/.jpeg/.gif) via print_image.
  • on_ai_message(message: Any, agent_name: str) -> None
    • Prints an AI message with agent-name color coding.
    • Extracts and displays <think>...</think> blocks as “Thoughts:” (and logs them).
    • Logs output with a fixed-width separator.
  • run_agent(agent: naas_abi_core.services.agent.Agent.Agent) -> None
    • Main interactive loop:
      • Initializes logging.
      • Registers agent hooks: tool usage, tool response, AI messages.
      • Prints a greeting ("{agent.name}: Hello, World!").
      • Repeatedly:
        • Shows active agent/model line (from agent.state.current_active_agent and agent list).
        • Reads user input.
        • Handles commands: /?, /reset, /bye, /exit, exit, quit.
        • Invokes agent.invoke(user_input).
        • Catches and logs exceptions + traceback.

Configuration/Dependencies​

  • Filesystem
    • Writes logs to: storage/datastore/interfaces/terminal_agent/<YYYYMMDDTHHMMSS>.txt
  • Terminal behavior
    • Uses termios/tty raw mode for interactive placeholder input (requires a TTY).
    • Uses ANSI escape sequences for cursor/line control.
  • Third-party libraries
    • rich (via console and rich.markdown.Markdown)
    • langchain_core.messages (AnyMessage, ToolMessage)
    • pydash (used to find the active agent by name)
  • Internal dependencies
    • naas_abi_core.apps.terminal_agent.terminal_style: console, print_image, print_tool_response, print_tool_usage
    • naas_abi_core.services.agent.Agent.Agent: agent interface with:
      • .on_tool_usage(...), .on_tool_response(...), .on_ai_message(...)
      • .invoke(text), .reset()
      • .agents, .name, .state.current_active_agent, .chat_model

Usage​

Minimal example (requires a valid Agent instance from your application):

from naas_abi_core.apps.terminal_agent.main import run_agent
from naas_abi_core.services.agent.Agent import Agent

agent = Agent(...)  # construct according to your project
run_agent(agent)

Interactive commands supported:

  • /? help
  • /reset reset the agent session
  • /bye, /exit, exit, quit end the session

Caveats​

  • Interactive placeholder input relies on raw terminal mode (termios/tty) and will not work on non-TTY stdin; in that case it falls back to plain input().
  • Conversation logging must be initialized (via init_conversation_file()), which run_agent() does automatically; otherwise save_to_conversation() is a no-op.
  • Image rendering is triggered by detecting space-separated “words” ending in .png/.jpg/.jpeg/.gif; paths with trailing punctuation may not be detected.