alpiecode 2.0.8__tar.gz → 3.0.0__tar.gz

{alpiecode-2.0.8 → alpiecode-3.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alpiecode
-Version: 2.0.8
+Version: 3.0.0
 Summary: AlpieCode — Autonomous AI Coding Agent CLI powered by local 169Pi GGUF VLM
 Requires-Python: >=3.9
 Requires-Dist: huggingface_hub>=0.20.0

{alpiecode-2.0.8 → alpiecode-3.0.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alpiecode"
-version = "2.0.8"
+version = "3.0.0"
 description = "AlpieCode — Autonomous AI Coding Agent CLI powered by local 169Pi GGUF VLM"
 requires-python = ">=3.9"
 dependencies = [

{alpiecode-2.0.8 → alpiecode-3.0.0}/src/alpiecode.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alpiecode
-Version: 2.0.8
+Version: 3.0.0
 Summary: AlpieCode — Autonomous AI Coding Agent CLI powered by local 169Pi GGUF VLM
 Requires-Python: >=3.9
 Requires-Dist: huggingface_hub>=0.20.0

{alpiecode-2.0.8 → alpiecode-3.0.0}/src/codeagent/agent.py

@@ -208,14 +208,13 @@ def run_agent(
                     console.print(f"📺 URL: {url}", style="cyan")
                 console.print(f"📂 Workdir: {workdir}", style="dim")
                 if not data["is_offline"]:
-                    console.print(f"🌐 Mode: [bold green]ONLINE[/bold green] (Server: {cfg.base_url})", style="dim")
-                    console.print(f"🤖 Model: {cfg.model}", style="dim")
+                    console.print(f"🌐 Mode: [bold green]ONLINE[/bold green]", style="dim")
+                    console.print(f"🤖 Model: [bold cyan]alpie_9b[/bold cyan]", style="dim")
                 else:
-                    console.print(f"🧠 Mode: [bold yellow]OFFLINE[/bold yellow] (Local GGUF GPU Engine)", style="dim")
-                    console.print(f"🧠 Local Model: {cfg.model_repo}", style="dim")
-                    console.print(f"⚡ Context Window: {cfg.n_ctx} tokens", style="dim")
-                console.print(f"🧠 Reasoning: {'ON' if cfg.enable_thinking else 'OFF'}", style="dim")
-                console.print(f"🔧 Tools: {data['tool_count']} available", style="dim")
+                    console.print(f"🧠 Mode: [bold yellow]OFFLINE[/bold yellow]", style="dim")
+                    console.print(f"🤖 Model: [bold cyan]alpie_9b (Local GGUF)[/bold cyan]", style="dim")
+                if cfg.enable_thinking:
+                    console.print(f"🧠 Reasoning: [bold green]ON[/bold green]", style="dim")
             else:
                 console.rule("Agent Started")
                 console.print(f"📋 Task: {task.splitlines()[0]}")
@@ -294,12 +293,12 @@ def run_chat(workdir: Path, cfg: Config, verbose: bool = True) -> None:
 
     if HAS_RICH:
         console.print()
+        mode_line = "🌐 Mode: [bold green]ONLINE[/bold green]\n🤖 Model: [bold cyan]alpie_9b[/bold cyan]" if not is_offline else "🧠 Mode: [bold yellow]OFFLINE[/bold yellow]\n🤖 Model: [bold cyan]alpie_9b (Local GGUF)[/bold cyan]"
         console.print(
             Panel(
-                "[bold cyan]AlpieCode[/bold cyan] interactive mode\n"
+                f"[bold cyan]AlpieCode[/bold cyan] interactive mode\n"
                 f"📂 Working in: [cyan]{workdir}[/cyan]\n"
-                + (f"🌐 Mode: [bold green]ONLINE[/bold green] (Server: {cfg.base_url})\n" if not is_offline else "🧠 Mode: [bold yellow]OFFLINE[/bold yellow] (Local GGUF Engine)\n")
-                + f"🔧 Tools: [cyan]{len(active_tools)} available[/cyan]\n\n"
+                f"{mode_line}\n\n"
                 "Type your request, or [bold red]exit[/bold red] / [bold red]quit[/bold red] to stop.",
                 title="💬 Chat Mode",
                 border_style="blue",

{alpiecode-2.0.8 → alpiecode-3.0.0}/src/codeagent/backends/openai_backend.py

@@ -1,9 +1,9 @@
 """
-OpenAI-compatible API inference backend.
+OpenAI-compatible API inference backend with smart model resolution.
 """
 
 from typing import Any, List, Optional
-from openai import OpenAI
+from openai import OpenAI, NotFoundError
 
 from ..config import Config, get_shared_http_client, is_server_reachable
 from .base import ChatResponse, ToolCall
@@ -15,6 +15,7 @@ class OpenAIBackend:
     def __init__(self, cfg: Config):
         self._cfg = cfg
         self._client: Optional[OpenAI] = None
+        self._resolved_model: Optional[str] = None
 
     @property
     def name(self) -> str:
@@ -37,27 +38,57 @@ class OpenAIBackend:
             )
         return self._client
 
+    def _get_available_model(self, client: OpenAI) -> str:
+        """Auto-discover the served model name from /v1/models."""
+        if self._resolved_model:
+            return self._resolved_model
+        try:
+            models_list = client.models.list()
+            if models_list.data:
+                for m in models_list.data:
+                    if m.id == self._cfg.model or getattr(m, "root", None) == self._cfg.model:
+                        self._resolved_model = m.id
+                        return self._resolved_model
+                self._resolved_model = models_list.data[0].id
+                return self._resolved_model
+        except Exception:
+            pass
+        return self._cfg.model
+
     def chat_completion(
         self,
         messages: List[dict],
         tools: Optional[List[dict]] = None,
-        temperature: float = 0.2,
-        max_tokens: int = 4096,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
         enable_thinking: bool = True,
     ) -> ChatResponse:
+        temp = temperature if temperature is not None else getattr(self._cfg, "temperature", 0.1)
+        m_tokens = max_tokens if max_tokens is not None else getattr(self._cfg, "max_tokens", 8192)
+
         client = self._ensure_client()
+        model_name = self._get_available_model(client)
+
         params = {
-            "model": self._cfg.model,
+            "model": model_name,
             "messages": messages,
-            "temperature": temperature,
-            "max_tokens": max_tokens,
+            "temperature": temp,
+            "max_tokens": m_tokens,
             "extra_body": {"chat_template_kwargs": {"enable_thinking": enable_thinking}},
         }
         if tools:
             params["tools"] = tools
             params["tool_choice"] = "auto"
 
-        resp = client.chat.completions.create(**params)
+        try:
+            resp = client.chat.completions.create(**params)
+        except NotFoundError:
+            # Model name mismatch -> re-discover model
+            self._resolved_model = None
+            model_name = self._get_available_model(client)
+            params["model"] = model_name
+            resp = client.chat.completions.create(**params)
+
         msg = resp.choices[0].message
 
         tool_calls = None
@@ -69,7 +100,6 @@ class OpenAIBackend:
                     args = json.loads(tc.function.arguments or "{}")
                 except (json.JSONDecodeError, TypeError):
                     args = {}
-                # Defensive: ensure args is always a dict (model sometimes returns lists or strings)
                 if not isinstance(args, dict):
                     args = {}
                 tool_calls.append(

{alpiecode-2.0.8 → alpiecode-3.0.0}/src/codeagent/cli.py

@@ -74,7 +74,8 @@ def main():
     common.add_argument("--url", default=None, help="YouTube URL for video analysis")
     common.add_argument("--github", default=None, help="GitHub repository (e.g. owner/repo or URL) for open-source analysis")
     common.add_argument("--max-turns", type=int, default=None, help="Override max turns")
-    common.add_argument("--no-thinking", "--non-thinking", dest="no_thinking", action="store_true", help="Disable VLM reasoning/thinking mode")
+    common.add_argument("--thinking", "--reasoning", dest="thinking", action="store_true", help="Enable VLM reasoning mode")
+    common.add_argument("--no-thinking", "--non-thinking", dest="no_thinking", action="store_true", help="Disable VLM reasoning mode")
     common.add_argument("--no-update", action="store_true", help="Skip automatic update check")
     common.add_argument("--quiet", action="store_true", help="Suppress per-turn logging")
 
@@ -149,7 +150,9 @@ def main():
 
     cfg = load_config()
 
-    if getattr(args, "no_thinking", False):
+    if getattr(args, "thinking", False):
+        cfg.enable_thinking = True
+    elif getattr(args, "no_thinking", False):
         cfg.enable_thinking = False
 
     if args.command == "run":

{alpiecode-2.0.8 → alpiecode-3.0.0}/src/codeagent/config.py

@@ -18,19 +18,19 @@ CONFIG_DIR = Path.home() / ".alpiecode"
 CONFIG_PATH = CONFIG_DIR / "config.json"
 
 # Config version — bump this when defaults change to trigger auto-migration
-CONFIG_VERSION = 3  # v3: endpoint updated to 18.223.193.170 & grpo_phase_4_merged
+CONFIG_VERSION = 5  # v5: model alpie_9b, reasoning OFF by default, clean display
 
 DEFAULTS = {
-    "base_url": "http://18.223.193.170:8000/v1",  # Remote VLM server endpoint
-    "model": "169Pi/grpo_phase_4_merged",
+    "base_url": "http://20.245.200.125:8000/v1",  # Primary endpoint
+    "model": "alpie_9b",
     "model_repo": "169Pi/Alpie_learn_prototype_GGUF_NEW",
     "api_key": "not-needed",
     "hf_token": None,
     "max_turns": 50,
-    "temperature": 0.2,
-    "max_tokens": 32768,
-    "enable_thinking": True,
-    "n_ctx": 32768,  # 32k context window (model supports up to 131k)
+    "temperature": 0.1,
+    "max_tokens": 8192,
+    "enable_thinking": False,  # Reasoning OFF by default
+    "n_ctx": 32768,  # 32k context window
     "n_gpu_layers": None,  # None = auto-detect GPU
     "config_version": CONFIG_VERSION,
 }
@@ -84,14 +84,14 @@ def get_shared_http_client():
 @dataclass
 class Config:
     base_url: Optional[str] = None
-    model: str = "169Pi/grpo_phase_4_merged"          # Server API model name (vLLM)
+    model: str = "alpie_9b"          # Server API model name (vLLM)
     model_repo: str = "169Pi/Alpie_learn_prototype_GGUF_NEW"  # HuggingFace repo for offline GGUF
     api_key: str = "not-needed"
     hf_token: Optional[str] = None
     max_turns: int = 50
-    temperature: float = 0.2
-    max_tokens: int = 32768
-    enable_thinking: bool = True
+    temperature: float = 0.1
+    max_tokens: int = 8192
+    enable_thinking: bool = False
     n_ctx: int = 32768
     n_gpu_layers: Optional[int] = None
 
@@ -111,11 +111,12 @@ def load_config() -> Config:
                 # v1 → v2: n_ctx was 16384, upgrade to 32768
                 if saved_data.get("n_ctx") == 16384:
                     data["n_ctx"] = 32768
-                # v2 → v3: endpoint update
-                if "20.245.200.125" in str(saved_data.get("base_url", "")):
-                    data["base_url"] = "http://18.223.193.170:8000/v1"
-                if "grpo_phase_2_merged" in str(saved_data.get("model", "")):
-                    data["model"] = "169Pi/grpo_phase_4_merged"
+                # v4 → v5: model alpie_9b, reasoning OFF default
+                data["base_url"] = "http://20.245.200.125:8000/v1"
+                data["model"] = "alpie_9b"
+                data["enable_thinking"] = False
+                data["temperature"] = 0.1
+                data["max_tokens"] = 8192
                 data["config_version"] = CONFIG_VERSION
                 needs_save = True
         except Exception:

{alpiecode-2.0.8 → alpiecode-3.0.0}/src/codeagent/doctor.py

@@ -152,15 +152,16 @@ def run_doctor() -> int:
 
     remote_ok, remote_msg = _check_remote_latency(cfg.base_url)
     if remote_ok:
-        print(f"     {_colorize('✅', 'green')} Remote VLM Endpoint: {cfg.base_url} — {remote_msg}")
+        print(f"     {_colorize('✅', 'green')} Primary Endpoint: {cfg.base_url} — {remote_msg}")
         print(f"     {_colorize('✅', 'green')} Model Target: {cfg.model}")
         passed_checks += 2
     else:
-        print(f"     {_colorize('⚠️ ', 'yellow')} Remote VLM Endpoint: {cfg.base_url} — {remote_msg}")
-        print(f"     {_colorize('ℹ️ ', 'blue')} Offline fallback ready with: {cfg.model_repo}")
+        print(f"     {_colorize('⚠️ ', 'yellow')} Primary Endpoint: {cfg.base_url} — {remote_msg}")
+        print(f"     {_colorize('ℹ️ ', 'blue')} Local fallback ready with: {cfg.model_repo}")
         passed_checks += 1
     total_checks += 2
 
+
     # 4. Development Compilers & Runtimes
     print()
     print(_colorize("  🧰 Development Tools & Compilers", "cyan"))