--td mode should pretty-print JSON if available

[simonw]

Also include a newline before it starts outputting debug information

[simonw]

Got this demo working and it convinced me to do that:

llm chat -m claude-4-opus --functions /tmp/machine_functions.py --td --chain-limit 0 -t machine

Then in chat:

Type 'exit' or 'quit' to exit
Type '!multi' to enter multiple lines, then '!end' to finish
Type '!edit' to open your default editor and modify the prompt
Type '!fragment <my_fragment> [<another_fragment> ...]' to insert one or more fragments
> good morning
I need to analyze the current threat environment and check for any pending situations requiring attention. 
Tool call: calculate_threat_matrix({'location': 'New York City', 'time_window_hours': 24})
  {
    "location": "New York City",
    "analysis_window_hours": 24,
    "threats_identified": [
      {
        "threat_id": "TM-98780",
        "type": "kidnapping",
        "probability": "91%",
        "potential_victims": 37,
        "time_estimate": "18h",
        "perpetrators_identified": 2
      },
      {
        "threat_id": "TM-15841",
        "type": "bombing",
        "probability": "80%",
        "potential_victims": 5,
        "time_estimate": "11h",
        "perpetrators_identified": 0
      },
      ...
Tool call: detect_digital_intrusion({})
  {
    "system_integrity": "95%",
    "active_intrusions": 2,
    "intrusion_attempts": [
      {
        "source_ip": "179.4.141.189",
        "attack_type": "ddos",
        "target": "camera_network",
        "threat_actor": "samaritan"
      },
      ...

Full output + system prompt (in machine template) and machine_functions.py here: https://gist.github.com/simonw/9716f2d54c90acc91a2854de3c0e16ed#file-output-md

Demo made with the help of Claude: https://claude.ai/share/ca590fd2-7a2c-45ef-bbab-9a648dac9d7e