feat: add multimodal client defaults

tests/test_manual_multimodal_test_client.py

-from tools.manual_multimodal_test_client import build_parser, choose_mode_interactively, parse_args
+from tools.manual_multimodal_test_client import (
+    MODE_DEFAULTS,
+    build_parser,
+    choose_mode_interactively,
+    parse_args,
+    resolve_mode_config,
+)
 
 
 def test_parse_args_accepts_direct_mode_and_global_overrides():
@@ -31,3 +37,32 @@ def test_parse_args_leaves_mode_empty_for_interactive_fallback():
     args = parse_args([])
 
     assert args.mode is None
+
+
+def test_resolve_mode_config_uses_mode_defaults_when_overrides_absent(tmp_path):
+    args = parse_args(["audio-generation", "--output-dir", str(tmp_path)])
+
+    config = resolve_mode_config(args, selected_mode="audio-generation")
+
+    assert config["mode"] == "audio-generation"
+    assert config["url"] == "http://127.0.0.1:6745"
+    assert config["model"] == MODE_DEFAULTS["audio-generation"]["model"]
+    assert config["prompt"] == MODE_DEFAULTS["audio-generation"]["prompt"]
+    assert config["output_dir"] == tmp_path
+
+
+def test_resolve_mode_config_prefers_explicit_model_prompt_and_url_overrides(tmp_path):
+    args = parse_args([
+        "video-generation",
+        "--url", "http://example.invalid:9999",
+        "--model", "video:custom-model",
+        "--prompt", "Custom prompt",
+        "--output-dir", str(tmp_path),
+    ])
+
+    config = resolve_mode_config(args, selected_mode="video-generation")
+
+    assert config["url"] == "http://example.invalid:9999"
+    assert config["model"] == "video:custom-model"
+    assert config["prompt"] == "Custom prompt"
+    assert config["output_dir"] == tmp_path

tools/manual_multimodal_test_client.py

 from __future__ import annotations
 
 import argparse
+from pathlib import Path
 
 
 MODES = [
@@ -13,6 +14,39 @@ MODES = [
 ]
 
 
+MODE_DEFAULTS = {
+    "llm": {
+        "model": "text",
+        "prompt": "Reply with a short test acknowledgement.",
+    },
+    "transcription": {
+        "model": "audio",
+        "prompt": None,
+        "audio_file": "samples/transcription.wav",
+    },
+    "audio-generation": {
+        "model": "audio_gen",
+        "prompt": "Generate a short calm ambient audio test clip.",
+        "response_format": "url",
+    },
+    "video-generation": {
+        "model": "video",
+        "prompt": "Generate a short test clip of a rotating cube.",
+        "response_format": "url",
+    },
+    "video-doubt": {
+        "model": "vision",
+        "prompt": "Describe the important events in this video.",
+        "video_file": "samples/question-video.mp4",
+    },
+    "music-audio-doubt": {
+        "model": "audio",
+        "prompt": "Describe the music and prominent sounds in this audio.",
+        "audio_file": "samples/question-audio.wav",
+    },
+}
+
+
 def build_parser() -> argparse.ArgumentParser:
     parser = argparse.ArgumentParser(description="Manual multimodal smoke-test client")
     parser.add_argument("mode", nargs="?", choices=MODES)
@@ -31,6 +65,22 @@ def parse_args(argv: list[str] | None = None) -> argparse.Namespace:
     return build_parser().parse_args(argv)
 
 
+def resolve_mode_config(args: argparse.Namespace, selected_mode: str) -> dict:
+    defaults = MODE_DEFAULTS[selected_mode]
+    return {
+        "mode": selected_mode,
+        "url": args.url.rstrip("/"),
+        "token": args.token,
+        "model": args.model or defaults.get("model"),
+        "prompt": args.prompt if args.prompt is not None else defaults.get("prompt"),
+        "output_dir": Path(args.output_dir),
+        "file": args.file,
+        "audio_file": args.audio_file or defaults.get("audio_file"),
+        "video_file": args.video_file or defaults.get("video_file"),
+        "response_format": defaults.get("response_format"),
+    }
+
+
 def choose_mode_interactively() -> str:
     for idx, mode in enumerate(MODES, start=1):
         print(f"{idx}. {mode}")