mlx-code 0.0.3__tar.gz → 0.0.4__tar.gz

{mlx_code-0.0.3 → mlx_code-0.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlx-code
-Version: 0.0.3
+Version: 0.0.4
 Summary: Coding Agent for Mac
 Home-page: https://github.com/JosefAlbers/mlx-code
 Author: J Joe
@@ -35,7 +35,7 @@ It features a multi-provider local server, a terminal-based chat REPL, and a ded
 ### Features
 
 *   **Local MLX Inference**: Powered by `mlx-lm` for optimized performance on Apple Silicon. Includes intelligent prompt caching.
-*   **Multi-Provider Compatibility**: Seamlessly translates and handles requests formatted for Claude, Gemini, Codex, and standard OpenAI APIs. 
+*   **Multi-Provider Compatibility**: Seamlessly translates and handles requests formatted for **Claude**, **Gemini**, **Codex**, **DeepSeek** and standard OpenAI APIs. 
 *   **Built-in REPL & Tools**: Comes with `pie`, a fully-featured chat REPL with tool execution and reasoning token support.
 *   **TUI Log Viewer**: Includes a Curses-based Terminal UI for filtering, inspecting, and tracking JSON logs in real-time.
 *   **Server Mode**: Easily spin up a local server compatible with standard LLM tooling.
@@ -70,8 +70,11 @@ mc --model mlx-community/Qwen3.5-4B-OptiQ-4bit
 # Use DeepSeek V4 Flash API
 me --deepseek
 
-# Run the server only on a custom port
-mc --nocc --port 8080
+# Run the server only
+mc --nocc
+
+# General shell piping and chaining works too
+echo "explain symgraph.py" | mc | cat - PLAN.md | mc
 ```
 *(For a full list of mc server arguments, run mc --help)*
 

{mlx_code-0.0.3 → mlx_code-0.0.4}/README.md

@@ -9,7 +9,7 @@ It features a multi-provider local server, a terminal-based chat REPL, and a ded
 ### Features
 
 *   **Local MLX Inference**: Powered by `mlx-lm` for optimized performance on Apple Silicon. Includes intelligent prompt caching.
-*   **Multi-Provider Compatibility**: Seamlessly translates and handles requests formatted for Claude, Gemini, Codex, and standard OpenAI APIs. 
+*   **Multi-Provider Compatibility**: Seamlessly translates and handles requests formatted for **Claude**, **Gemini**, **Codex**, **DeepSeek** and standard OpenAI APIs. 
 *   **Built-in REPL & Tools**: Comes with `pie`, a fully-featured chat REPL with tool execution and reasoning token support.
 *   **TUI Log Viewer**: Includes a Curses-based Terminal UI for filtering, inspecting, and tracking JSON logs in real-time.
 *   **Server Mode**: Easily spin up a local server compatible with standard LLM tooling.
@@ -44,8 +44,11 @@ mc --model mlx-community/Qwen3.5-4B-OptiQ-4bit
 # Use DeepSeek V4 Flash API
 me --deepseek
 
-# Run the server only on a custom port
-mc --nocc --port 8080
+# Run the server only
+mc --nocc
+
+# General shell piping and chaining works too
+echo "explain symgraph.py" | mc | cat - PLAN.md | mc
 ```
 *(For a full list of mc server arguments, run mc --help)*
 

{mlx_code-0.0.3 → mlx_code-0.0.4}/mlx_code/main.py

@@ -335,7 +335,7 @@ def parse_default(body: dict) -> tuple[list[Tool], list[Message]]:
         for b in content:
             btype = b.get("type")
             if btype == "thinking":
-                thinking_parts.append(b.get("thinking") or b.get("text", ""))
+                thinking_parts.append(b.get("thinking") or b.get("reasoning_content") or b.get("text", ""))
             elif btype == "text":
                 text_parts.append(b.get("text", ""))
         return "\n".join(text_parts), "\n".join(thinking_parts)
@@ -355,7 +355,7 @@ def parse_default(body: dict) -> tuple[list[Tool], list[Message]]:
             messages.append(Message(
                 role="assistant",
                 content=text or None,
-                thinking=thinking or m.get("thinking") or None,
+                thinking=thinking or m.get("reasoning_content") or m.get("thinking") or None,
                 tool_calls=[
                     ToolCall(tc["id"], tc["function"]["name"], _safe_json(tc["function"]["arguments"]))
                     for tc in m["tool_calls"]
@@ -367,7 +367,7 @@ def parse_default(body: dict) -> tuple[list[Tool], list[Message]]:
             messages.append(Message(
                 role=role,
                 content=text or None,
-                thinking=thinking or m.get("thinking") or None,
+                thinking=thinking or m.get("reasoning_content") or m.get("thinking") or None,
             ))
 
     return tools, messages
@@ -1111,7 +1111,7 @@ class DefaultAdapter(BaseAdapter):
         if not text:
             return b""
         if state == "thinking":
-            return self.chunk({"thinking": text})
+            return self.chunk({"reasoning_content": text})
         return self.chunk({"content": text})
 
     def tool(self, tool):

{mlx_code-0.0.3 → mlx_code-0.0.4}/mlx_code/pie.py

@@ -516,7 +516,7 @@ class DefaultChat:
                 if text_parts:
                     msg["content"] = "".join(b.text for b in text_parts)
                 if thinking_parts:
-                    msg["thinking"] = "".join(b.thinking for b in thinking_parts)
+                    msg["reasoning_content"] = "".join(b.thinking for b in thinking_parts)
                 if tool_calls:
                     msg["tool_calls"] = [
                         {"id": tc.id, "type": "function",
@@ -585,7 +585,7 @@ class DefaultChat:
             msg = AssistantMessage()
             try:
                 async with httpx.AsyncClient(timeout=120.0) as client:
-                    async with client.stream("POST", f"{self.base_url}/chat/completions", json=payload, headers=headers) as resp:
+                    async with client.stream("POST", f"{self.base_url}/v1/chat/completions", json=payload, headers=headers) as resp:
                         if resp.status_code >= 400:
                             body = await resp.aread()
                             raise RuntimeError(f"HTTP {resp.status_code}: {body.decode()}")
@@ -610,10 +610,10 @@ class DefaultChat:
                             delta = choice.get("delta", {})
                             finish_reason = choice.get("finish_reason") or finish_reason
 
-                            if delta.get("thinking"):
-                                text = delta["thinking"]
-                                _thinking_buf += text
-                                es.push(Event("thinking_delta", {"delta": text, "partial": msg}))
+                            reasoning = delta.get("reasoning_content") or delta.get("thinking")
+                            if reasoning:
+                                _thinking_buf += reasoning
+                                es.push(Event("thinking_delta", {"delta": reasoning, "partial": msg}))
 
                             if delta.get("content"):
                                 text = delta["content"]
@@ -1986,11 +1986,18 @@ async def _repl(
     agent = Agent(api, system=system, tools=available_tools)
     loop = asyncio.get_running_loop()
     _suppress = False
+    last_ev_type = ""
+    last_delta = ""
     async def on_event(event: AgentEvent) -> None:
         # logger.debug(event) # □
-        nonlocal _suppress, is_tty
+        nonlocal _suppress, is_tty, last_ev_type, last_delta
         if event.type == "text_delta":
             delta = event.payload.get("delta", "")
+
+            if last_ev_type and last_ev_type[:4] != event.type[:4] and last_delta and not last_delta[-1].isspace() and delta and not delta[0].isspace():
+                print()
+            last_delta = delta
+            last_ev_type = event.type
             if "<tool_call>" in delta:
                 before, _, _ = delta.partition("<tool_call>")
                 print(before.strip(), end="", flush=True)
@@ -2004,10 +2011,19 @@ async def _repl(
         elif is_tty:
             if event.type == "thinking_delta":
                 delta = event.payload.get("delta", "")
+                if last_ev_type and last_ev_type[:4] != event.type[:4] and last_delta and not last_delta[-1].isspace() and delta and not delta[0].isspace():
+                    print()
+                last_delta = delta
+                last_ev_type = event.type
                 if delta.strip():
                     print(f"\033[2m{delta}\033[0m", end="", flush=True)
             elif event.type == "tool_start":
-                print(f"\033[33m{event.payload['name']}:\033[0m {json.dumps(event.payload['args'])[:120]}\n", flush=True)
+                delta = f"\033[33m{event.payload['name']}:\033[0m {json.dumps(event.payload['args'])[:120]}\n"
+                if last_ev_type and last_ev_type[:4] != event.type[:4] and last_delta and not last_delta[-1].isspace() and delta and not delta[0].isspace():
+                    print()
+                last_delta = delta
+                last_ev_type = event.type
+                print(delta, flush=True)
             elif event.type == "tool_result":
                 msg = event.payload["message"]
                 raw = "\n".join(b.text for b in msg.content if isinstance(b, TextContent))
@@ -2015,10 +2031,10 @@ async def _repl(
                 # print(f"\n\n\033[36m{raw[:200]}\033[0m\n", end="", flush=True) # □
             elif event.type == "tool_end":
                 if event.payload.get("is_error"):
-                    print(" \033[31m(error)\033[0m", end="", flush=True)
+                    print(" \033[31m(error)\033[0m", flush=True)
             elif event.type == "error":
                 err = event.payload.get("error")
-                print(f"\n\033[31m[error]\033[0m {getattr(err, 'error_message', str(err))}")
+                print(f"\n\033[31m[error]\033[0m {getattr(err, 'error_message', str(err))}\n")
 
     agent.subscribe(on_event)
     if is_tty:
@@ -2037,6 +2053,8 @@ async def _repl(
         if not user_input:
             continue
 
+        last_delta = ""
+        last_ev_type = ""
         logger.info(user_input)
 
         if user_input.startswith("/"):
@@ -2108,7 +2126,7 @@ def run_repl(
     elif provider == "codex":
         api = CodexChat(model=model, api_key=os.environ.get("OPENAI_API_KEY") if api_key is None else api_key, base_url=f'{base_url}/v1' if base_url else "https://api.openai.com/v1")
     else:
-        api = DefaultChat(model=model, api_key="mp" if api_key is None else api_key, base_url=f'{base_url}/v1' if base_url else "https://api.openai.com/v1")
+        api = DefaultChat(model=model, api_key="mp" if api_key is None else api_key, base_url=base_url if base_url else "https://api.openai.com/v1")
     
     try:
         asyncio.run(_repl(api, system=system, cwd=cwd, tools=tools))
@@ -2132,7 +2150,7 @@ def main():
     if args.simulate:
         asyncio.run(simulate())
     elif args.deepseek:
-        run_repl(base_url="https://api.deepseek.com/anthropic", model="deepseek-v4-flash", provider="claude", api_key=os.environ.get('DEEPSEEK_API_KEY'))
+        run_repl(base_url="https://api.deepseek.com", model="deepseek-v4-flash", provider="default", api_key=os.environ.get('DEEPSEEK_API_KEY'))
     else:
         run_repl(
             model=args.model, 

{mlx_code-0.0.3 → mlx_code-0.0.4}/mlx_code.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlx-code
-Version: 0.0.3
+Version: 0.0.4
 Summary: Coding Agent for Mac
 Home-page: https://github.com/JosefAlbers/mlx-code
 Author: J Joe
@@ -35,7 +35,7 @@ It features a multi-provider local server, a terminal-based chat REPL, and a ded
 ### Features
 
 *   **Local MLX Inference**: Powered by `mlx-lm` for optimized performance on Apple Silicon. Includes intelligent prompt caching.
-*   **Multi-Provider Compatibility**: Seamlessly translates and handles requests formatted for Claude, Gemini, Codex, and standard OpenAI APIs. 
+*   **Multi-Provider Compatibility**: Seamlessly translates and handles requests formatted for **Claude**, **Gemini**, **Codex**, **DeepSeek** and standard OpenAI APIs. 
 *   **Built-in REPL & Tools**: Comes with `pie`, a fully-featured chat REPL with tool execution and reasoning token support.
 *   **TUI Log Viewer**: Includes a Curses-based Terminal UI for filtering, inspecting, and tracking JSON logs in real-time.
 *   **Server Mode**: Easily spin up a local server compatible with standard LLM tooling.
@@ -70,8 +70,11 @@ mc --model mlx-community/Qwen3.5-4B-OptiQ-4bit
 # Use DeepSeek V4 Flash API
 me --deepseek
 
-# Run the server only on a custom port
-mc --nocc --port 8080
+# Run the server only
+mc --nocc
+
+# General shell piping and chaining works too
+echo "explain symgraph.py" | mc | cat - PLAN.md | mc
 ```
 *(For a full list of mc server arguments, run mc --help)*
 

{mlx_code-0.0.3 → mlx_code-0.0.4}/setup.py

@@ -6,7 +6,7 @@ setup(
     author_email="albersj66@gmail.com",
     author="J Joe",
     license="Apache-2.0",
-    version="0.0.3",
+    version="0.0.4",
     readme="README.md",
     description="Coding Agent for Mac",
     long_description=open("README.md").read(),