Initial deployment of Claude Code with Qwen local

Dockerfile

@@ -0,0 +1,51 @@
+FROM python:3.12-slim
+
+# Install system dependencies
+RUN apt-get update && apt-get install -y \
+    git \
+    curl \
+    build-essential \
+    libopenblas-dev \
+    nodejs \
+    npm \
+    && rm -rf /var/lib/apt/lists/*
+
+# Set working directory
+WORKDIR /app
+
+# Install llama-cpp-python with OpenBLAS and Streamlit
+RUN pip install llama-cpp-python[server] streamlit
+
+# Download Qwen2.5-Coder-7B-Instruct GGUF Q4_K_M
+RUN mkdir -p /app/models
+RUN curl -L -o /app/models/qwen2.5-coder-7b-instruct-q4_k_m.gguf https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct-GGUF/resolve/main/qwen2.5-coder-7b-instruct-q4_k_m.gguf
+
+# Install Claude Code globally
+RUN npm install -g @anthropic-ai/claude-code
+
+# Create a dummy config.toml for OpenManus if needed, or just for structure
+RUN mkdir -p /app/config
+RUN echo '[llm]\nmodel = "qwen2.5-coder"\nbase_url = "http://localhost:8001/v1"\napi_key = "not-needed"\nmax_tokens = 4096\ntemperature = 0.0\n' > /app/config/config.toml
+
+# Copy our custom GUI for Claude Code
+COPY app_claude_code_gui.py /app/app_claude_code_gui.py
+
+# Create a startup script
+RUN echo '#!/bin/bash\n\
+# Start llama-cpp-python server in background\n\
+python3 -m llama_cpp.server --model /app/models/qwen2.5-coder-7b-instruct-q4_k_m.gguf --host 0.0.0.0 --port 8001 &\n\
+\n\
+# Wait for server to start\n\
+until curl -s http://localhost:8001/v1/models > /dev/null; do\n\
+  echo "Waiting for LLM server..."\n\
+  sleep 5\n\
+done\n\
+\n\
+# Start Streamlit UI on port 7860\n\
+streamlit run app_claude_code_gui.py --server.port 7860 --server.address 0.0.0.0\n\
+' > /app/start.sh && chmod +x /app/start.sh
+
+# HF Spaces expect a web service on port 7860
+EXPOSE 7860
+
+CMD ["/app/start.sh"]

app_claude_code_gui.py

@@ -0,0 +1,81 @@
+import streamlit as st
+import subprocess
+import threading
+import queue
+import os
+import sys
+
+st.set_page_config(page_title="Claude Code Web UI", page_icon="💻")
+
+st.title("💻 Claude Code Web UI")
+st.markdown("Interface pour Claude Code avec Qwen2.5-Coder-7B (GGUF) local")
+
+if "messages" not in st.session_state:
+    st.session_state.messages = []
+
+for message in st.session_state.messages:
+    with st.chat_message(message["role"]):
+        st.markdown(message["content"])
+
+def run_command(command, output_queue):
+    try:
+        # Ensure claude-code uses the local LLM server
+        # This might require setting environment variables or a config file for claude-code
+        # For now, we assume claude-code can be configured to use an OpenAI-compatible endpoint
+        # or we might need to modify its source/config if it's hardcoded to Anthropic API.
+        # Let's try to pass the API base URL as an environment variable if claude-code supports it.
+        # If not, this part will need further investigation.
+        
+        # For demonstration, we'll just run a simple command or simulate claude-code interaction.
+        # A real integration would involve piping input/output to the claude-code CLI.
+        
+        # Example: running a simple shell command for now
+        process = subprocess.Popen(
+            command, 
+            shell=True, 
+            stdout=subprocess.PIPE, 
+            stderr=subprocess.STDOUT, 
+            text=True, 
+            bufsize=1
+        )
+        for line in iter(process.stdout.readline, ""):
+            output_queue.put(line)
+        process.stdout.close()
+        process.wait()
+        output_queue.put(None)
+    except Exception as e:
+        output_queue.put(f"Error executing command: {e}\n")
+        output_queue.put(None)
+
+if prompt := st.chat_input("Entrez votre commande Claude Code ou shell..."):
+    st.session_state.messages.append({"role": "user", "content": prompt})
+    with st.chat_message("user"):
+        st.markdown(prompt)
+
+    with st.chat_message("assistant"):
+        response_placeholder = st.empty()
+        full_response = ""
+        output_queue = queue.Queue()
+        
+        # Prepend 'claude-code' if the user just types a command, or allow direct shell commands
+        command_to_run = prompt
+        if not prompt.startswith("claude-code") and not prompt.startswith("npm") and not prompt.startswith("python") and not prompt.startswith("ls"):
+            command_to_run = f"claude-code {prompt}"
+
+        thread = threading.Thread(target=run_command, args=([command_to_run], output_queue))
+        thread.start()
+        
+        while True:
+            try:
+                line = output_queue.get(timeout=1)
+                if line is None:
+                    break
+                full_response += line
+                response_placeholder.markdown(f"```bash\n{full_response}▌\n```")
+            except queue.Empty:
+                if not thread.is_alive():
+                    break
+                continue
+        
+        response_placeholder.markdown(f"```bash\n{full_response}\n```")
+        st.session_state.messages.append({"role": "assistant", "content": full_response})