Merge pull request #87 from B-AROL-O/feat/mcp-pupper

feat: make FREISA-GPT main code public

Author: gmacario

.devcontainer/Dockerfile

@@ -1,4 +1,4 @@
-FROM ghcr.io/tiryoh/ros2-desktop-vnc:humble-20250831T0421
+FROM ghcr.io/tiryoh/ros2-desktop-vnc:humble
 
 ARG USERNAME
 

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "code/mcp-client-pupper/ros-mcp-server"]
+	path = code/mcp-client-pupper/ros-mcp-server
+	url = https://github.com/davmacario/ros-mcp-server.git

code/FREISA-GPT/.flake8

@@ -0,0 +1,3 @@
+[flake8]
+max-line-length = 120
+extend-ignore = E203,E701

code/FREISA-GPT/.python-version

@@ -0,0 +1 @@
+3.13

code/FREISA-GPT/README.md

@@ -0,0 +1,74 @@
+import asyncio
+import os
+from argparse import ArgumentParser
+from pathlib import Path
+
+from dotenv import load_dotenv
+
+from src.mcp_client_pupper.main import main
+from src.mcp_client_pupper.puppy_voice_assistant import PuppyVoiceAssistant
+
+BASE_URL = "https://open-webui.dmhosted.duckdns.org"
+DEFAULT_MODEL = "gpt-oss:20b"
+ROSBRIDGE_ADDRESS = "ws://localhost:9090"
+DEFAULT_PUPPY_API_URL = "http://localhost:5080"
+
+if __name__ == "__main__":
+    load_dotenv()
+    parser = ArgumentParser()
+    parser.add_argument(
+        "--llm-base-url",
+        type=str,
+        help="Base URL to reach the LLM (without path)",
+        default=os.getenv("OPENAI_BASE_URL", BASE_URL),
+    )
+    parser.add_argument(
+        "--llm-model",
+        type=str,
+        help="LLM model. To be chosen among the ones available at the specified `--llm-base-url`",
+        default=os.getenv("OPENAI_MODEL", DEFAULT_MODEL),
+    )
+    parser.add_argument(
+        "--mcp-server-config",
+        type=Path,
+        help="Path to the MCP server configuration JSON file.",
+        default=Path("./servers_config.json"),
+    )
+    parser.add_argument(
+        "--puppy-api-url",
+        type=str,
+        help="URL of the HTTP API for the Puppy State API",
+        default=os.getenv("PUPPY_API_URL", DEFAULT_PUPPY_API_URL),
+    )
+    parser.add_argument(
+        "--input-device",
+        type=int,
+        default=1,
+        help=f"Id of The input device (aka microphone)\navailable devices {PuppyVoiceAssistant.available_devices()}",
+    )
+    parser.add_argument(
+        "--whisper-model",
+        default="small.en",
+        type=str,
+        help="Whisper.cpp model, default to %(default)s",
+    )
+    parser.add_argument(
+        "--silence-threshold",
+        default=16,
+        type=int,
+        help="The duration of silence after which the inference will be running, defaults to %(default)s",
+    )
+    parser.add_argument(
+        "--block-duration",
+        default=30,
+        help="Minimum time audio updates in ms, default to %(default)s",
+    )
+    parser.add_argument(  # TODO: remove; it is only used in MCP Server
+        "--rosbridge-address",
+        type=str,
+        help="Address of the websocket on which rosbridge is exposed. Format: 'ws://<address>:<port>'",
+        default=os.getenv("ROSBRIDGE_ADDRESS", ROSBRIDGE_ADDRESS),
+    )
+    args = parser.parse_args()
+
+    asyncio.run(main(args))

code/FREISA-GPT/main.py

@@ -0,0 +1,39 @@
+[project]
+name = "FREISA-GPT"
+version = "0.1.0"
+description = ""
+readme = "README.md"
+requires-python = ">=3.13"
+dependencies = [
+    "fastapi-mcp>=0.2.0",
+    "httpx>=0.28.1",
+    "linkup-sdk>=0.2.4",
+    "fastmcp>=2.11.3",
+    "mcp[cli]>=1.13.0",
+    "opencv-python>=4.11.0.86",
+    "pillow>=11.3.0",
+    "websocket-client>=1.8.0",
+    "flask>=3.1.2",
+    "icecream>=2.1.5",
+    "marimo>=0.14.16",
+    "openai>=1.99.5",
+    "python-dotenv>=1.1.1",
+    "pywhispercpp",
+    "setuptools>=80.9.0",
+    "sounddevice>=0.5.2",
+    "soundfile>=0.13.1",
+    "webrtcvad>=2.0.10",
+]
+
+[dependency-groups]
+dev = [
+    "pyright>=1.1.405",
+    "ruff>=0.13.0",
+]
+
+[tool.ruff]
+line-length = 120
+indent-width = 4
+
+[tool.ruff.lint]
+extend-select = ["I"]

code/FREISA-GPT/pyproject.toml

@@ -0,0 +1,15 @@
+{
+  "mcpServers": {
+    "ros-mcp-server": {
+      "command": "uv",
+      "args": [
+        "--directory",
+        "./src/mcp_server_pupper",
+        "run",
+        "main.py",
+        "--mcp-transport",
+        "stdio"
+      ]
+    }
+  }
+}

code/FREISA-GPT/servers_config.json

@@ -0,0 +1,35 @@
+import logging
+import os
+
+from .mcp_client import ChatSessionConfig
+from .puppy_voice_assistant import PuppyVoiceAssistant, VoiceConfig
+from .utils.llm_client import LLMClientConfig
+
+"""
+FREISA-GPT entrypoint
+"""
+
+
+logging.basicConfig(
+    level=os.getenv("FREISA_LOG_LEVEL", "DEBUG"),
+    format="%(asctime)s [%(levelname)-8s] [%(name)s]: %(message)s",
+    datefmt="%Y-%m-%d %H:%M:%S",
+)
+
+
+async def main(args):
+    api_key = os.getenv("OPENAI_API_KEY")
+
+    chat_config = ChatSessionConfig(
+        mcp_server_config_file=args.mcp_server_config,
+        llm_client_config=LLMClientConfig(base_url=args.llm_base_url, model=args.llm_model, api_key=api_key),
+    )
+    voice_config = VoiceConfig(
+        model=args.whisper_model,
+        input_device=args.input_device,
+        silence_threshold=args.silence_threshold,
+        block_duration=args.block_duration,
+    )
+    voice_assistant = PuppyVoiceAssistant(voice_config, chat_config, args.puppy_api_url)
+    await voice_assistant.set_up_mcp_client()
+    voice_assistant.start()