augint-shell 0.76.0__tar.gz → 0.77.0__tar.gz

{augint_shell-0.76.0 → augint_shell-0.77.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: augint-shell
-Version: 0.76.0
+Version: 0.77.0
 Summary: Launch AI coding tools and local LLMs in Docker containers
 Author: svange
 Requires-Dist: docker>=7.0.0
@@ -106,7 +106,7 @@ extra_env = { MY_VAR = "value" }
 
 [llm]
 primary_model = "qwen3-coder:32b-a3b-q4_K_M"
-fallback_model = "qwen3.5:27b"
+fallback_model = "huihui_ai/llama3.3-abliterated"
 context_size = 32768
 ollama_port = 11434
 webui_port = 3000

{augint_shell-0.76.0 → augint_shell-0.77.0}/README.md

@@ -93,7 +93,7 @@ extra_env = { MY_VAR = "value" }
 
 [llm]
 primary_model = "qwen3-coder:32b-a3b-q4_K_M"
-fallback_model = "qwen3.5:27b"
+fallback_model = "huihui_ai/llama3.3-abliterated"
 context_size = 32768
 ollama_port = 11434
 webui_port = 3000

{augint_shell-0.76.0 → augint_shell-0.77.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "augint-shell"
-version = "0.76.0"
+version = "0.77.0"
 description = "Launch AI coding tools and local LLMs in Docker containers"
 authors = [{name = "svange"}]
 readme = "README.md"

{augint_shell-0.76.0 → augint_shell-0.77.0}/src/ai_shell/__init__.py

@@ -1,6 +1,6 @@
 """augint-shell (ai-shell) - Launch AI coding tools and local LLMs in Docker containers."""
 
-__version__ = "0.76.0"
+__version__ = "0.77.0"
 
 __all__ = [
     "__version__",

{augint_shell-0.76.0 → augint_shell-0.77.0}/src/ai_shell/cli/commands/llm.py

@@ -1,5 +1,6 @@
 """LLM stack management commands: up, down, pull, setup, status, logs, shell."""
 
+import socket
 import time
 from pathlib import Path
 
@@ -9,7 +10,7 @@ from rich.console import Console
 from ai_shell.cli import CONTEXT_SETTINGS
 from ai_shell.config import load_config
 from ai_shell.container import ContainerManager
-from ai_shell.defaults import OLLAMA_CONTAINER, WEBUI_CONTAINER
+from ai_shell.defaults import LOBECHAT_CONTAINER, OLLAMA_CONTAINER, WEBUI_CONTAINER
 from ai_shell.gpu import get_vram_info, get_vram_processes
 
 console = Console(stderr=True)
@@ -17,6 +18,26 @@ console = Console(stderr=True)
 _LOW_MEMORY_THRESHOLD_GIB = 30  # 27B+ models need ~30 GiB
 
 
+def _lan_ip() -> str | None:
+    """Return the host's primary LAN IPv4 address, or None if undetectable.
+
+    Uses a UDP socket's routing-table selection without actually sending
+    traffic. Works on Linux, Mac, and WSL2. On WSL2 this returns the
+    WSL VM's eth0 address (typically 172.x.x.x), which is reachable from
+    the Windows host but not the broader LAN unless WSL mirrored mode or
+    a Windows portproxy is configured.
+    """
+    try:
+        with socket.socket(socket.AF_INET, socket.SOCK_DGRAM) as s:
+            s.connect(("8.8.8.8", 80))
+            ip = str(s.getsockname()[0])
+    except OSError:
+        return None
+    if ip.startswith("127."):
+        return None
+    return ip
+
+
 def _warn_if_low_memory() -> None:
     """Check system memory and warn if it may be insufficient for large models."""
     try:
@@ -59,13 +80,13 @@ def _get_manager(ctx) -> ContainerManager:
 @click.group("llm", context_settings=CONTEXT_SETTINGS)
 @click.pass_context
 def llm_group(ctx):
-    """Manage the local LLM stack (Ollama + Open WebUI)."""
+    """Manage the local LLM stack (Ollama + Open WebUI + LobeChat)."""
 
 
 @llm_group.command("up")
 @click.pass_context
 def llm_up(ctx):
-    """Start the LLM stack (Ollama + Open WebUI)."""
+    """Start the LLM stack (Ollama + Open WebUI + LobeChat)."""
     manager = _get_manager(ctx)
     console.print("[bold]Starting LLM stack...[/bold]")
     _warn_if_low_memory()
@@ -76,6 +97,18 @@ def llm_up(ctx):
     manager.ensure_webui()
     console.print(f"  Open WebUI:  http://localhost:{manager.config.webui_port}")
 
+    manager.ensure_lobechat()
+    console.print(
+        f"  LobeChat:    http://localhost:{manager.config.lobechat_port}  [dim](recommended)[/dim]"
+    )
+
+    lan = _lan_ip()
+    if lan:
+        console.print("\n[bold]LAN access[/bold] (bound to 0.0.0.0):")
+        console.print(f"  Ollama API:  http://{lan}:{manager.config.ollama_port}")
+        console.print(f"  Open WebUI:  http://{lan}:{manager.config.webui_port}")
+        console.print(f"  LobeChat:    http://{lan}:{manager.config.lobechat_port}")
+
     console.print("\n[bold green]LLM stack is running.[/bold green]")
     console.print("If this is your first time, run: [bold]ai-shell llm setup[/bold]")
 
@@ -87,7 +120,7 @@ def llm_down(ctx):
     manager = _get_manager(ctx)
     console.print("[bold]Stopping LLM stack...[/bold]")
 
-    for name in [WEBUI_CONTAINER, OLLAMA_CONTAINER]:
+    for name in [LOBECHAT_CONTAINER, WEBUI_CONTAINER, OLLAMA_CONTAINER]:
         status = manager.container_status(name)
         if status == "running":
             manager.stop_container(name)
@@ -132,6 +165,7 @@ def llm_setup(ctx):
     _warn_if_low_memory()
     manager.ensure_ollama()
     manager.ensure_webui()
+    manager.ensure_lobechat()
 
     # Wait for Ollama to be ready
     console.print("[bold]Waiting for Ollama to be ready...[/bold]")
@@ -157,23 +191,10 @@ def llm_setup(ctx):
     output = manager.exec_in_ollama(["ollama", "pull", config.fallback_model])
     console.print(output)
 
-    # Configure context window
-    console.print(f"\n[bold]Configuring context window ({config.context_size} tokens)...[/bold]")
-    for model in [config.primary_model, config.fallback_model]:
-        modelfile = f"FROM {model}\nPARAMETER num_ctx {config.context_size}\n"
-        # Write modelfile and create model
-        manager.exec_in_ollama(
-            [
-                "sh",
-                "-c",
-                f'printf "{modelfile}" > /tmp/Modelfile && '
-                f"ollama create {model} -f /tmp/Modelfile && rm -f /tmp/Modelfile",
-            ]
-        )
-
     console.print("\n[bold green]============================================[/bold green]")
     console.print("[bold green] Setup complete![/bold green]")
-    console.print(f"\n  Open WebUI:  http://localhost:{config.webui_port}")
+    console.print(f"\n  LobeChat:    http://localhost:{config.lobechat_port}  (recommended)")
+    console.print(f"  Open WebUI:  http://localhost:{config.webui_port}")
     console.print(f"  Ollama API:  http://localhost:{config.ollama_port}")
     console.print(f"\n  Primary model:  {config.primary_model}")
     console.print(f"  Fallback model: {config.fallback_model}")
@@ -189,9 +210,14 @@ def llm_status(ctx):
     config = manager.config
     ollama_running = manager.container_status(OLLAMA_CONTAINER) == "running"
     webui_running = manager.container_status(WEBUI_CONTAINER) == "running"
+    lobechat_running = manager.container_status(LOBECHAT_CONTAINER) == "running"
 
     console.print("[bold]Container status:[/bold]")
-    for name, running in [(OLLAMA_CONTAINER, ollama_running), (WEBUI_CONTAINER, webui_running)]:
+    for name, running in [
+        (OLLAMA_CONTAINER, ollama_running),
+        (WEBUI_CONTAINER, webui_running),
+        (LOBECHAT_CONTAINER, lobechat_running),
+    ]:
         status = manager.container_status(name)
         if running:
             console.print(f"  {name}: [green]{status}[/green]")
@@ -203,6 +229,13 @@ def llm_status(ctx):
     console.print("\n[bold]Access URLs:[/bold]")
     ollama_url = f"http://localhost:{config.ollama_port}"
     webui_url = f"http://localhost:{config.webui_port}"
+    lobechat_url = f"http://localhost:{config.lobechat_port}"
+    if lobechat_running:
+        console.print(
+            f"  LobeChat:           [cyan]{lobechat_url}[/cyan]  [bold](recommended)[/bold]"
+        )
+    else:
+        console.print(f"  LobeChat:           [dim]{lobechat_url}[/dim] (not running)")
     if ollama_running:
         console.print(f"  Ollama API:         [cyan]{ollama_url}[/cyan]")
         console.print(f"  OpenAI-compatible:  [cyan]{ollama_url}/v1[/cyan]")
@@ -211,10 +244,15 @@ def llm_status(ctx):
         console.print(f"  OpenAI-compatible:  [dim]{ollama_url}/v1[/dim] (not running)")
     if webui_running:
         console.print(f"  Open WebUI:         [cyan]{webui_url}[/cyan]")
-        console.print(f"  Chat interface:     [cyan]{webui_url}[/cyan]")
     else:
         console.print(f"  Open WebUI:         [dim]{webui_url}[/dim] (not running)")
-        console.print(f"  Chat interface:     [dim]{webui_url}[/dim] (not running)")
+
+    lan = _lan_ip()
+    if lan:
+        console.print("\n[bold]LAN access[/bold] (bound to 0.0.0.0):")
+        console.print(f"  Ollama API:         [cyan]http://{lan}:{config.ollama_port}[/cyan]")
+        console.print(f"  Open WebUI:         [cyan]http://{lan}:{config.webui_port}[/cyan]")
+        console.print(f"  LobeChat:           [cyan]http://{lan}:{config.lobechat_port}[/cyan]")
 
     console.print("\n[bold]Configuration:[/bold]")
     console.print(f"  Primary model:   {config.primary_model}")
@@ -252,7 +290,7 @@ def llm_logs(ctx, follow):
     if follow:
         manager.container_logs(OLLAMA_CONTAINER, follow=True)
     else:
-        for name in [OLLAMA_CONTAINER, WEBUI_CONTAINER]:
+        for name in [OLLAMA_CONTAINER, WEBUI_CONTAINER, LOBECHAT_CONTAINER]:
             status = manager.container_status(name)
             if status is not None:
                 console.print(f"\n[bold]--- {name} ---[/bold]")

{augint_shell-0.76.0 → augint_shell-0.77.0}/src/ai_shell/config.py

@@ -26,6 +26,7 @@ from ai_shell.defaults import (
     DEFAULT_DEV_PORTS,
     DEFAULT_FALLBACK_MODEL,
     DEFAULT_IMAGE,
+    DEFAULT_LOBECHAT_PORT,
     DEFAULT_OLLAMA_PORT,
     DEFAULT_PRIMARY_MODEL,
     DEFAULT_WEBUI_PORT,
@@ -50,6 +51,7 @@ class AiShellConfig:
     context_size: int = DEFAULT_CONTEXT_SIZE
     ollama_port: int = DEFAULT_OLLAMA_PORT
     webui_port: int = DEFAULT_WEBUI_PORT
+    lobechat_port: int = DEFAULT_LOBECHAT_PORT
 
     # Extra configuration
     extra_env: dict[str, str] = field(default_factory=dict)
@@ -180,6 +182,8 @@ def _apply_config(config: AiShellConfig, path: Path) -> None:
         config.ollama_port = int(llm["ollama_port"])
     if "webui_port" in llm:
         config.webui_port = int(llm["webui_port"])
+    if "lobechat_port" in llm:
+        config.lobechat_port = int(llm["lobechat_port"])
 
     # [aws] section
     aws = data.get("aws", {})
@@ -213,6 +217,7 @@ def _apply_env_vars(config: AiShellConfig) -> None:
         "AI_SHELL_CONTEXT_SIZE": ("context_size", int),
         "AI_SHELL_OLLAMA_PORT": ("ollama_port", int),
         "AI_SHELL_WEBUI_PORT": ("webui_port", int),
+        "AI_SHELL_LOBECHAT_PORT": ("lobechat_port", int),
         "AI_SHELL_AI_PROFILE": ("ai_profile", str),
         "AI_SHELL_AWS_REGION": ("aws_region", str),
         "AI_SHELL_BEDROCK_PROFILE": ("bedrock_profile", str),

{augint_shell-0.76.0 → augint_shell-0.77.0}/src/ai_shell/container.py

@@ -19,6 +19,8 @@ from docker.types import DeviceRequest, Mount
 import docker
 from ai_shell.defaults import (
     LLM_NETWORK,
+    LOBECHAT_CONTAINER,
+    LOBECHAT_IMAGE,
     OLLAMA_CONTAINER,
     OLLAMA_CPU_SHARES,
     OLLAMA_DATA_VOLUME,
@@ -274,7 +276,9 @@ class ContainerManager:
         # GPU auto-detection
         gpu_available = detect_gpu()
         device_requests = None
-        env: dict[str, str] = {}
+        env: dict[str, str] = {
+            "OLLAMA_CONTEXT_LENGTH": str(self.config.context_size),
+        }
         if gpu_available:
             device_requests = [DeviceRequest(count=1, capabilities=[["gpu"]])]
             vram = get_vram_info()
@@ -358,6 +362,42 @@ class ContainerManager:
         logger.info("Open WebUI container created on port %d", self.config.webui_port)
         return WEBUI_CONTAINER
 
+    def ensure_lobechat(self) -> str:
+        """Get or create the LobeChat container.
+
+        Client-DB mode: state lives in the browser's IndexedDB, so no
+        persistent server-side volume is needed.
+
+        Returns the container name.
+        """
+        container = self._get_container(LOBECHAT_CONTAINER)
+
+        if container is not None:
+            if container.status != "running":
+                logger.info("Starting existing LobeChat container")
+                container.start()
+            return LOBECHAT_CONTAINER
+
+        logger.info("Creating LobeChat container")
+        self._pull_image_if_needed(LOBECHAT_IMAGE)
+        network_name = self._ensure_llm_network()
+
+        self.client.containers.run(
+            image=LOBECHAT_IMAGE,
+            name=LOBECHAT_CONTAINER,
+            ports={"3210/tcp": ("0.0.0.0", self.config.lobechat_port)},  # nosec B104
+            environment={
+                "OLLAMA_PROXY_URL": f"http://{OLLAMA_CONTAINER}:11434/v1",
+                "ACCESS_CODE": "",
+            },
+            restart_policy={"Name": "unless-stopped"},
+            detach=True,
+            network=network_name,
+        )
+
+        logger.info("LobeChat container created on port %d", self.config.lobechat_port)
+        return LOBECHAT_CONTAINER
+
     def exec_in_ollama(self, command: list[str]) -> str:
         """Run a command in the Ollama container and return stdout.
 

{augint_shell-0.76.0 → augint_shell-0.77.0}/src/ai_shell/defaults.py

@@ -53,11 +53,13 @@ WEBUI_DATA_VOLUME = "augint-shell-webui-data"
 # =============================================================================
 OLLAMA_IMAGE = "ollama/ollama"
 WEBUI_IMAGE = "ghcr.io/open-webui/open-webui:main"
+LOBECHAT_IMAGE = "lobehub/lobe-chat:latest"
 DEFAULT_PRIMARY_MODEL = "qwen3-coder:32b-a3b-q4_K_M"
-DEFAULT_FALLBACK_MODEL = "qwen3.5:27b"
+DEFAULT_FALLBACK_MODEL = "huihui_ai/llama3.3-abliterated"
 DEFAULT_CONTEXT_SIZE = 32768
 DEFAULT_OLLAMA_PORT = 11434
 DEFAULT_WEBUI_PORT = 3000
+DEFAULT_LOBECHAT_PORT = 3210
 DEFAULT_DEV_PORTS = [3000, 4200, 5000, 5173, 5678, 8000, 8080, 8888]
 
 # =============================================================================
@@ -76,6 +78,7 @@ OLLAMA_CPU_SHARES = 1024  # Docker CPU scheduling priority (default 0 = fair-sha
 # =============================================================================
 OLLAMA_CONTAINER = "augint-shell-ollama"
 WEBUI_CONTAINER = "augint-shell-webui"
+LOBECHAT_CONTAINER = "augint-shell-lobechat"
 
 # =============================================================================
 # Docker network

{augint_shell-0.76.0 → augint_shell-0.77.0}/src/ai_shell/templates/ai-shell.yaml

@@ -157,7 +157,7 @@
 #
 # llm:
 #   primary_model: qwen3-coder:32b-a3b-q4_K_M
-#   fallback_model: qwen3.5:27b
+#   fallback_model: huihui_ai/llama3.3-abliterated
 #   context_size: 32768
 #   ollama_port: 11434
 #   webui_port: 3000