yaicli 0.6.2__py3-none-any.whl → 0.6.4__py3-none-any.whl

pyproject.toml

@@ -1,10 +1,10 @@
 [project]
 name = "yaicli"
-version = "0.6.2"
+version = "0.6.4"
 description = "A simple CLI tool to interact with LLM"
 authors = [{ name = "belingud", email = "im.victor@qq.com" }]
 readme = "README.md"
-requires-python = ">=3.9"
+requires-python = ">=3.10"
 license = { file = "LICENSE" }
 classifiers = [
     "Programming Language :: Python :: 3",
@@ -42,6 +42,15 @@ keywords = [
     "anthropic",
     "groq",
     "cohere",
+    "huggingface",
+    "chatglm",
+    "sambanova",
+    "siliconflow",
+    "xai",
+    "vertexai",
+    "deepseek",
+    "modelscope",
+    "ollama",
 ]
 dependencies = [
     "click>=8.1.8",
@@ -65,10 +74,20 @@ ai = "yaicli.entry:app"
 yaicli = "yaicli.entry:app"
 
 [project.optional-dependencies]
+all = [
+    "volcengine-python-sdk>=3.0.15",
+    "ollama>=0.5.1",
+    "cohere>=5.15.0",
+    "google-genai>=1.20.0",
+    "huggingface-hub>=0.33.0",
+]
 doubao = ["volcengine-python-sdk>=3.0.15"]
 ollama = ["ollama>=0.5.1"]
 cohere = ["cohere>=5.15.0"]
-all = ["volcengine-python-sdk>=3.0.15", "ollama>=0.5.1", "cohere>=5.15.0"]
+gemini = ["google-genai>=1.20.0"]
+huggingface = [
+    "huggingface-hub>=0.33.0",
+]
 
 [tool.pytest.ini_options]
 testpaths = ["tests"]
@@ -91,6 +110,7 @@ dev = [
     "pytest>=8.3.5",
     "pytest-cov>=6.1.1",
     "ruff>=0.11.2",
+    "tox>=4.27.0",
 ]
 
 [tool.isort]

yaicli/cli.py

@@ -267,7 +267,7 @@ class CLI:
                     assistant_msg = self.chat.history[i + 1] if (i + 1) < len(self.chat.history) else None
                     self.console.print(f"[dim]{i // 2 + 1}[/dim] [bold blue]User:[/bold blue] {user_msg.content}")
                     if assistant_msg:
-                        md = Markdown(assistant_msg.content, code_theme=cfg["CODE_THEME"])
+                        md = Markdown(assistant_msg.content or "", code_theme=cfg["CODE_THEME"])
                         padded_md = Padding(md, (0, 0, 0, 4))
                         self.console.print("    Assistant:", style="bold green")
                         self.console.print(padded_md)
@@ -384,14 +384,7 @@ class CLI:
         self._check_history_len()
 
         if self.current_mode == EXEC_MODE:
-            # We need to extract the executable command from the last assistant message
-            # in case of tool use.
-            final_content = ""
-            if self.chat.history:
-                last_message = self.chat.history[-1]
-                if last_message.role == "assistant":
-                    final_content = last_message.content or ""
-            self._confirm_and_execute(final_content)
+            self._confirm_and_execute(content or "")
         return True
 
     def _confirm_and_execute(self, raw_content: str) -> None:

yaicli/config.py

@@ -142,7 +142,7 @@ class Config(dict):
                 if target_type is bool:
                     converted_value = str2bool(raw_value)
                 elif target_type in (int, float, str):
-                    converted_value = target_type(raw_value)
+                    converted_value = target_type(raw_value) if raw_value else raw_value
                 elif target_type is dict and raw_value:
                     converted_value = json.loads(raw_value)
             except (ValueError, TypeError, json.JSONDecodeError) as e:

yaicli/const.py

@@ -1,5 +1,5 @@
 try:
-    from enum import StrEnum
+    from enum import StrEnum  # type: ignore
 except ImportError:
     from enum import Enum
 

yaicli/llms/__init__.py

@@ -1,13 +1,4 @@
-from ..config import cfg
 from .client import LLMClient
 from .provider import Provider, ProviderFactory
 
 __all__ = ["LLMClient", "Provider", "ProviderFactory"]
-
-
-class BaseProvider:
-    def __init__(self) -> None:
-        self.api_key = cfg["API_KEY"]
-        self.model = cfg["MODEL"]
-        self.base_url = cfg["BASE_URL"]
-        self.timeout = cfg["TIMEOUT"]

yaicli/llms/provider.py

@@ -43,9 +43,13 @@ class ProviderFactory:
         "chatglm": (".providers.chatglm_provider", "ChatglmProvider"),
         "chutes": (".providers.chutes_provider", "ChutesProvider"),
         "cohere": (".providers.cohere_provider", "CohereProvider"),
+        "cohere-bedrock": (".providers.cohere_provider", "CohereBadrockProvider"),
+        "cohere-sagemaker": (".providers.cohere_provider", "CohereSagemakerProvider"),
         "deepseek": (".providers.deepseek_provider", "DeepSeekProvider"),
         "doubao": (".providers.doubao_provider", "DoubaoProvider"),
+        "gemini": (".providers.gemini_provider", "GeminiProvider"),
         "groq": (".providers.groq_provider", "GroqProvider"),
+        "huggingface": (".providers.huggingface_provider", "HuggingFaceProvider"),
         "infini-ai": (".providers.infiniai_provider", "InfiniAIProvider"),
         "minimax": (".providers.minimax_provider", "MinimaxProvider"),
         "modelscope": (".providers.modelscope_provider", "ModelScopeProvider"),
@@ -55,6 +59,7 @@ class ProviderFactory:
         "sambanova": (".providers.sambanova_provider", "SambanovaProvider"),
         "siliconflow": (".providers.siliconflow_provider", "SiliconFlowProvider"),
         "targon": (".providers.targon_provider", "TargonProvider"),
+        "vertexai": (".providers.vertexai_provider", "VertexAIProvider"),
         "xai": (".providers.xai_provider", "XaiProvider"),
         "yi": (".providers.yi_provider", "YiProvider"),
     }

yaicli/llms/providers/chatglm_provider.py

@@ -1,5 +1,5 @@
 import json
-from typing import Any, Dict, Generator, Optional
+from typing import Generator, Optional
 
 from openai._streaming import Stream
 from openai.types.chat.chat_completion import ChatCompletion, Choice
@@ -14,10 +14,14 @@ class ChatglmProvider(OpenAIProvider):
 
     DEFAULT_BASE_URL = "https://open.bigmodel.cn/api/paas/v4/"
 
-    def get_completion_params(self) -> Dict[str, Any]:
-        params = super().get_completion_params()
-        params["max_tokens"] = params.pop("max_completion_tokens")
-        return params
+    COMPLETION_PARAMS_KEYS = {
+        "model": "MODEL",
+        "temperature": "TEMPERATURE",
+        "top_p": "TOP_P",
+        "max_tokens": "MAX_TOKENS",
+        "do_sample": "DO_SAMPLE",
+        "extra_body": "EXTRA_BODY",
+    }
 
     def _handle_normal_response(self, response: ChatCompletion) -> Generator[LLMResponse, None, None]:
         """Handle normal (non-streaming) response

yaicli/llms/providers/cohere_provider.py

@@ -10,7 +10,8 @@ This module implements Cohere provider classes for different deployment options:
 from typing import Any, Dict, Generator, List, Optional
 
 from cohere import BedrockClientV2, ClientV2, SagemakerClientV2
-from cohere.types.tool_call_v2 import ToolCallV2, ToolCallV2Function
+from cohere.types.tool_call_v2 import ToolCallV2
+from cohere.types.tool_call_v2function import ToolCallV2Function
 
 from ...config import cfg
 from ...console import get_console
@@ -179,7 +180,9 @@ class CohereProvider(Provider):
                 continue
             elif chunk.type == "tool-call-delta":
                 # Tool call arguments being generated: cohere.types.chat_tool_call_delta_event_delta_message.ChatToolCallDeltaEventDeltaMessage
-                tool_call.arguments += chunk.delta.message.tool_calls.function.arguments
+                if not tool_call:
+                    continue
+                tool_call.arguments += chunk.delta.message.tool_calls.function.arguments or ""
                 # Waiting for tool-call-end event
                 continue
 
@@ -292,7 +295,7 @@ class CohereBadrockProvider(CohereProvider):
         return self.CLIENT_CLS(**self.client_params)
 
 
-class CohereSagemaker(CohereBadrockProvider):
+class CohereSagemakerProvider(CohereBadrockProvider):
     """Cohere provider for AWS Sagemaker integration"""
 
     CLIENT_CLS = SagemakerClientV2

yaicli/llms/providers/deepseek_provider.py

@@ -10,5 +10,6 @@ class DeepSeekProvider(OpenAIProvider):
 
     def get_completion_params(self) -> Dict[str, Any]:
         params = super().get_completion_params()
-        params["max_tokens"] = params.pop("max_completion_tokens")
+        if "max_completion_tokens" in params:
+            params["max_tokens"] = params.pop("max_completion_tokens")
         return params

yaicli/llms/providers/gemini_provider.py

@@ -0,0 +1,193 @@
+import json
+from functools import wraps
+from typing import Any, Callable, Dict, Generator, List
+
+import google.genai as genai
+from google.genai import types
+
+from ...config import cfg
+from ...console import get_console
+from ...schemas import ChatMessage, LLMResponse
+from ...tools import get_func_name_map
+from ..provider import Provider
+
+
+def wrap_function(func):
+    @wraps(func)
+    def wrapper(*args, **kwargs):
+        return func(*args, **kwargs)
+
+    return wrapper
+
+
+class GeminiProvider(Provider):
+    """Gemini provider implementation based on google-genai library"""
+
+    DEFAULT_BASE_URL = "https://generativelanguage.googleapis.com/v1beta"
+
+    def __init__(self, config: dict = cfg, verbose: bool = False, **kwargs):
+        self.config = config
+        self.enable_function = self.config["ENABLE_FUNCTIONS"]
+        self.verbose = verbose
+
+        # Initialize client
+        self.client_params = self.get_client_params()
+        self.client = genai.Client(**self.client_params)
+        self.console = get_console()
+
+    def get_client_params(self) -> Dict[str, Any]:
+        """Get the client parameters"""
+        # Initialize client params
+        return {
+            "api_key": self.config["API_KEY"],
+        }
+
+    def get_chat_config(self):
+        http_options_map = {
+            "timeout": self.config["TIMEOUT"] * 1000,  # Timeout for the request in milliseconds.
+            "headers": {**self.config["EXTRA_HEADERS"], "X-Client": self.APP_NAME, "Referer": self.APP_REFERER},
+        }
+        if self.config.get("BASE_URL"):
+            http_options_map["base_url"] = self.config["BASE_URL"]
+        if self.config.get("API_VERSION"):
+            # Specifies the version of the API to use.
+            http_options_map["api_version"] = self.config["API_VERSION"]
+        http_options = types.HttpOptions(**http_options_map)
+        config_map = {
+            "max_output_tokens": self.config["MAX_TOKENS"],
+            "temperature": self.config["TEMPERATURE"],
+            "top_p": self.config["TOP_P"],
+            "http_options": http_options,
+        }
+        if self.config.get("TOP_K"):
+            config_map["top_k"] = self.config["TOP_K"]
+        if self.config.get("PRESENCE_PENALTY"):
+            config_map["presence_penalty"] = self.config["PRESENCE_PENALTY"]
+        if self.config.get("FREQUENCY_PENALTY"):
+            config_map["frequency_penalty"] = self.config["FREQUENCY_PENALTY"]
+        if self.config.get("SEED"):
+            config_map["seed"] = self.config["SEED"]
+        # Indicates whether to include thoughts in the response. If true, thoughts are returned only if the model supports thought and thoughts are available.
+        thinking_config_map = {"include_thoughts": self.config.get("INCLUDE_THOUGHTS", True)}
+        if self.config.get("THINKING_BUDGET"):
+            thinking_config_map["thinking_budget"] = int(self.config["THINKING_BUDGET"])
+        config_map["thinking_config"] = types.ThinkingConfig(**thinking_config_map)
+        config = types.GenerateContentConfig(**config_map)
+        if self.enable_function:
+            # TODO: support disable automatic function calling
+            # config.automatic_function_calling = types.AutomaticFunctionCallingConfig(disable=False)
+            config.tools = self.gen_gemini_functions()
+        return config
+
+    def _convert_messages(self, messages: List[ChatMessage]) -> List[types.Content]:
+        """Convert a list of ChatMessage objects to a list of Gemini Content objects."""
+        converted_messages = []
+        for msg in messages:
+            if msg.role == "system":
+                continue
+            content = types.Content(role=self._map_role(msg.role), parts=[types.Part(text=msg.content)])
+            if msg.role == "tool":
+                content.role = "user"
+                content.parts = [
+                    types.Part.from_function_response(name=msg.name or "", response={"result": msg.content})
+                ]
+            converted_messages.append(content)
+        return converted_messages
+
+    def _map_role(self, role: str) -> str:
+        """Map OpenAI roles to Gemini roles"""
+        # Gemini uses "user", "model" instead of "user", "assistant"
+        if role == "assistant":
+            return "model"
+        return role
+
+    def gen_gemini_functions(self) -> List[Callable[..., Any]]:
+        """Wrap Gemini functions from OpenAI functions for automatic function calling"""
+        func_name_map = get_func_name_map()
+        if not func_name_map:
+            return []
+        funcs = []
+        for func_name, func in func_name_map.items():
+            wrapped_func = wrap_function(func.execute)
+            wrapped_func.__name__ = func_name
+            wrapped_func.__doc__ = func.__doc__
+            funcs.append(wrapped_func)
+        return funcs
+
+    def completion(
+        self,
+        messages: List[ChatMessage],
+        stream: bool = False,
+    ) -> Generator[LLMResponse, None, None]:
+        """
+        Send completion request to Gemini and return responses.
+
+        Args:
+            messages: List of chat messages to send
+            stream: Whether to stream the response
+
+        Yields:
+            LLMResponse: Response objects containing content, tool calls, etc.
+
+        Raises:
+            ValueError: If messages is empty or invalid
+            APIError: If API request fails
+        """
+        gemini_messages = self._convert_messages(messages)
+        if self.verbose:
+            self.console.print("Messages:")
+            self.console.print(gemini_messages)
+        chat_config = self.get_chat_config()
+        chat_config.system_instruction = messages[0].content
+        chat = self.client.chats.create(model=self.config["MODEL"], history=gemini_messages, config=chat_config)  # type: ignore
+        message = messages[-1].content
+
+        if stream:
+            response = chat.send_message_stream(message=message)  # type: ignore
+            yield from self._handle_stream_response(response)
+        else:
+            response = chat.send_message(message=message)  # type: ignore
+            yield from self._handle_normal_response(response)
+
+    def _handle_normal_response(self, response) -> Generator[LLMResponse, None, None]:
+        """Handle normal (non-streaming) response"""
+        # TODO: support disable automatic function calling
+        if not response or not response.candidates:
+            yield LLMResponse(
+                content=json.dumps(response.to_json_dict()),
+                finish_reason="stop",
+            )
+            return
+        for part in response.candidates[0].content.parts:
+            if part.thought:
+                yield LLMResponse(reasoning=part.text, finish_reason="stop")
+            else:
+                yield LLMResponse(reasoning=None, content=part.text, finish_reason="stop")
+
+    def _handle_stream_response(self, response) -> Generator[LLMResponse, None, None]:
+        """Handle streaming response from Gemini API"""
+        # Initialize tool call object to accumulate tool call data across chunks
+        # TODO: support disable automatic function calling
+        tool_call = None
+        for chunk in response:
+            if not chunk.candidates:
+                continue
+            candidate = chunk.candidates[0]
+            finish_reason = candidate.finish_reason
+            for part in chunk.candidates[0].content.parts:
+                if part.thought:
+                    reasoning = part.text
+                    content = None
+                else:
+                    content = part.text
+                    reasoning = None
+                yield LLMResponse(
+                    reasoning=reasoning,
+                    content=content or "",
+                    tool_call=tool_call if finish_reason == "tool_calls" else None,
+                    finish_reason=finish_reason or None,
+                )
+
+    def detect_tool_role(self) -> str:
+        """Return the role that should be used for tool responses"""
+        return "user"

yaicli/llms/providers/huggingface_provider.py

@@ -0,0 +1,40 @@
+from typing import Any, Dict
+
+from huggingface_hub import InferenceClient
+
+from .chatglm_provider import ChatglmProvider
+
+
+class HuggingFaceProvider(ChatglmProvider):
+    """
+    HuggingFaceProvider is a provider for the HuggingFace API.
+    """
+
+    CLIENT_CLS = InferenceClient
+    DEFAULT_PROVIDER = "hf-inference"
+
+    COMPLETION_PARAMS_KEYS = {
+        "model": "MODEL",
+        "temperature": "TEMPERATURE",
+        "top_p": "TOP_P",
+        "max_tokens": "MAX_TOKENS",
+        "extra_body": "EXTRA_BODY",
+    }
+
+    def get_client_params(self) -> Dict[str, Any]:
+        client_params = {
+            "api_key": self.config["API_KEY"],
+            "timeout": self.config["TIMEOUT"],
+            "provider": self.config.get("HF_PROVIDER") or self.DEFAULT_PROVIDER,
+        }
+        if self.config["BASE_URL"]:
+            client_params["base_url"] = self.config["BASE_URL"]
+        if self.config["EXTRA_HEADERS"]:
+            client_params["headers"] = {
+                **self.config["EXTRA_HEADERS"],
+                "X-Title": self.APP_NAME,
+                "HTTP-Referer": self.APP_REFERER,
+            }
+        if self.config.get("BILL_TO"):
+            client_params["bill_to"] = self.config["BILL_TO"]
+        return client_params

yaicli/llms/providers/infiniai_provider.py

@@ -1,5 +1,6 @@
 from typing import Any, Dict
 
+from ...config import cfg
 from .openai_provider import OpenAIProvider
 
 
@@ -8,7 +9,7 @@ class InfiniAIProvider(OpenAIProvider):
 
     DEFAULT_BASE_URL = "https://cloud.infini-ai.com/maas/v1"
 
-    def __init__(self, config: dict = ..., **kwargs):
+    def __init__(self, config: dict = cfg, **kwargs):
         super().__init__(config, **kwargs)
         if self.enable_function:
             self.console.print("InfiniAI does not support functions, disabled", style="yellow")
@@ -16,5 +17,6 @@ class InfiniAIProvider(OpenAIProvider):
 
     def get_completion_params(self) -> Dict[str, Any]:
         params = super().get_completion_params()
-        params["max_tokens"] = params.pop("max_completion_tokens")
+        if "max_completion_tokens" in params:
+            params["max_tokens"] = params.pop("max_completion_tokens")
         return params

yaicli/llms/providers/modelscope_provider.py

@@ -10,5 +10,6 @@ class ModelScopeProvider(OpenAIProvider):
 
     def get_completion_params(self) -> Dict[str, Any]:
         params = super().get_completion_params()
-        params["max_tokens"] = params.pop("max_completion_tokens")
+        if "max_completion_tokens" in params:
+            params["max_tokens"] = params.pop("max_completion_tokens")
         return params

yaicli/llms/providers/openai_provider.py

@@ -19,7 +19,7 @@ class OpenAIProvider(Provider):
     DEFAULT_BASE_URL = "https://api.openai.com/v1"
     CLIENT_CLS = openai.OpenAI
     # Base mapping between config keys and API parameter names
-    _BASE_COMPLETION_PARAMS_KEYS = {
+    COMPLETION_PARAMS_KEYS = {
         "model": "MODEL",
         "temperature": "TEMPERATURE",
         "top_p": "TOP_P",
@@ -69,7 +69,7 @@ class OpenAIProvider(Provider):
         Returns:
             Dict[str, str]: Mapping from API parameter names to config keys
         """
-        return self._BASE_COMPLETION_PARAMS_KEYS.copy()
+        return self.COMPLETION_PARAMS_KEYS.copy()
 
     def get_completion_params(self) -> Dict[str, Any]:
         """
@@ -81,7 +81,7 @@ class OpenAIProvider(Provider):
         completion_params = {}
         params_keys = self.get_completion_params_keys()
         for api_key, config_key in params_keys.items():
-            if self.config.get(config_key, None) is not None:
+            if self.config.get(config_key, None) is not None and self.config[config_key] != "":
                 completion_params[api_key] = self.config[config_key]
         return completion_params
 
@@ -89,7 +89,7 @@ class OpenAIProvider(Provider):
         """Convert a list of ChatMessage objects to a list of OpenAI message dicts."""
         converted_messages = []
         for msg in messages:
-            message = {"role": msg.role, "content": msg.content or ""}
+            message: Dict[str, Any] = {"role": msg.role, "content": msg.content or ""}
 
             if msg.name:
                 message["name"] = msg.name
@@ -140,12 +140,19 @@ class OpenAIProvider(Provider):
             if tools:
                 params["tools"] = tools
 
-        if stream:
-            response = self.client.chat.completions.create(**params)
-            yield from self._handle_stream_response(response)
-        else:
-            response = self.client.chat.completions.create(**params)
-            yield from self._handle_normal_response(response)
+        try:
+            if stream:
+                response = self.client.chat.completions.create(**params)
+                yield from self._handle_stream_response(response)
+            else:
+                response = self.client.chat.completions.create(**params)
+                yield from self._handle_normal_response(response)
+        except (openai.APIStatusError, openai.APIResponseValidationError) as e:
+            try:
+                body = e.response.json()
+            except Exception:
+                body = e.response.text
+            self.console.print(f"Error Response: {body}")
 
     def _handle_normal_response(self, response: ChatCompletion) -> Generator[LLMResponse, None, None]:
         """Handle normal (non-streaming) response"""

yaicli/llms/providers/siliconflow_provider.py

@@ -10,5 +10,6 @@ class SiliconFlowProvider(OpenAIProvider):
 
     def get_completion_params(self) -> Dict[str, Any]:
         params = super().get_completion_params()
-        params["max_tokens"] = params.pop("max_completion_tokens")
+        if "max_completion_tokens" in params:
+            params["max_tokens"] = params.pop("max_completion_tokens")
         return params

yaicli/llms/providers/vertexai_provider.py

@@ -0,0 +1,18 @@
+from typing import Any, Dict
+
+from .gemini_provider import GeminiProvider
+
+
+class VertexAIProvider(GeminiProvider):
+    """Vertex AI provider implementation based on google-genai library"""
+
+    def get_client_params(self) -> Dict[str, Any]:
+        """Get the client parameters"""
+        # Initialize client params
+        if not self.config.get("PROJECT") or not self.config.get("LOCATION"):
+            raise ValueError("PROJECT and LOCATION are required for Vertex AI")
+        return {
+            "vertexai": True,
+            "project": self.config.get("PROJECT"),
+            "location": self.config.get("LOCATION"),
+        }

{yaicli-0.6.2.dist-info → yaicli-0.6.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: yaicli
-Version: 0.6.2
+Version: 0.6.4
 Summary: A simple CLI tool to interact with LLM
 Project-URL: Homepage, https://github.com/belingud/yaicli
 Project-URL: Repository, https://github.com/belingud/yaicli
@@ -208,11 +208,11 @@ License:                                  Apache License
            See the License for the specific language governing permissions and
            limitations under the License.
 License-File: LICENSE
-Keywords: ai,ai-assistant,ai-chat,ai-interaction,anthropic,chatgpt,claude,cli,cohere,command-line,completion,console-application,conversation,gemini,gpt,groq,inference,interactive,language-model,llm,llms,mistral,nlp,openai,prompt,python-tool,shell-integration,terminal,terminal-interface,text-generation
+Keywords: ai,ai-assistant,ai-chat,ai-interaction,anthropic,chatglm,chatgpt,claude,cli,cohere,command-line,completion,console-application,conversation,deepseek,gemini,gpt,groq,huggingface,inference,interactive,language-model,llm,llms,mistral,modelscope,nlp,ollama,openai,prompt,python-tool,sambanova,shell-integration,siliconflow,terminal,terminal-interface,text-generation,vertexai,xai
 Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python :: 3
-Requires-Python: >=3.9
+Requires-Python: >=3.10
 Requires-Dist: click>=8.1.8
 Requires-Dist: distro>=1.9.0
 Requires-Dist: httpx>=0.28.1
@@ -225,12 +225,18 @@ Requires-Dist: socksio>=1.0.0
 Requires-Dist: typer>=0.16.0
 Provides-Extra: all
 Requires-Dist: cohere>=5.15.0; extra == 'all'
+Requires-Dist: google-genai>=1.20.0; extra == 'all'
+Requires-Dist: huggingface-hub>=0.33.0; extra == 'all'
 Requires-Dist: ollama>=0.5.1; extra == 'all'
 Requires-Dist: volcengine-python-sdk>=3.0.15; extra == 'all'
 Provides-Extra: cohere
 Requires-Dist: cohere>=5.15.0; extra == 'cohere'
 Provides-Extra: doubao
 Requires-Dist: volcengine-python-sdk>=3.0.15; extra == 'doubao'
+Provides-Extra: gemini
+Requires-Dist: google-genai>=1.20.0; extra == 'gemini'
+Provides-Extra: huggingface
+Requires-Dist: huggingface-hub>=0.33.0; extra == 'huggingface'
 Provides-Extra: ollama
 Requires-Dist: ollama>=0.5.1; extra == 'ollama'
 Description-Content-Type: text/markdown
@@ -323,14 +329,8 @@ Yaicli has several optional dependencies group, you can copy below commands to i
 # install all denpendencies
 pip install 'yaicli[all]'
 
-# install with ollama support
-pip instsall 'yaicli[ollama]'
-
-# install with cohere support
-pip install 'yaicli[cohere]'
-
-# install with doubao support
-pip install 'yaicli[doubao]'
+# install with specific provider support
+pip instsall 'yaicli[ollama,cohere,doubao,huggingface,gemini]'
 ```
 
 Install by `uv`.
@@ -339,14 +339,8 @@ Install by `uv`.
 # install all denpendencies
 uv tool install 'yaicli[all]'
 
-# install with ollama support
-uv tool instsall 'yaicli[ollama]'
-
-# install with cohere support
-uv tool install 'yaicli[cohere]'
-
-# install with doubao support
-uv tool install 'yaicli[doubao]'
+# install with specific provider support
+uv tool instsall 'yaicli[ollama,cohere,doubao,huggingface,gemini]'
 ```
 
 ### Install from Source
@@ -357,6 +351,31 @@ cd yaicli
 pip install .
 ```
 
+## Buildin Supported Providers
+
+- AI21
+- Chatglm
+- Chuts
+- Cohere
+- Cohere Badrock
+- Cohere Sagemaker
+- Deepseek
+- Doubao
+- Gemini
+- Vertex AI
+- Groq
+- Huggingface
+- Minimax
+- ModelScope
+- Ollama
+- Openai
+- Sambanova
+- Siliconflow
+- Targon
+- X AI
+- Yi
+- Unlimited OpenAI-compatible providers
+
 ## ⚙️ Configuration
 
 YAICLI uses a simple configuration file to store your preferences and API keys.
@@ -494,6 +513,15 @@ API_KEY=
 MODEL=gpt-4o
 ```
 
+Extra params:
+
+```ini
+# REASONING_EFFORT: [high, midium, low]
+REASONING_EFFORT=
+```
+
+See official for more details: https://platform.openai.com/docs/guides/reasoning?api-mode=chat
+
 #### Deepseek
 
 ```ini
@@ -510,6 +538,48 @@ API_KEY=
 MODEL=deepseek/deepseek-chat-v3-0324
 ```
 
+#### Gemini
+
+Basic config:
+
+```ini
+PROVIDER=gemini
+API_KEY=
+MODEL=gemini-2.5-flash
+```
+
+Extra params:
+
+```ini
+TOP_K=
+PRESENCE_PENALTY=
+FREQUENCY_PENALTY=
+SEED=
+THINKING_BUDGET=
+API_VERSION=
+BASE_URL=
+```
+
+#### Vertex AI
+
+```ini
+PROVIDER=vertexai
+MODEL=gemini-2.5-flash
+PROJECT=
+LOCATION=
+```
+
+#### Huggingface
+
+```ini
+HF_PROVIDER=sambanova
+PROVIDER=huggingface
+API_KEY=
+MODEL=deepseek-ai/DeepSeek-R1-0528
+```
+
+See official docs for `HF_PROVIDER`: https://huggingface.co/docs/inference-providers/index
+
 #### Groq
 
 ```ini
@@ -534,6 +604,15 @@ API_KEY=
 MODEL=glm-4-plus
 ```
 
+Extra params:
+
+Check offcial docs: https://bigmodel.cn/dev/api/normal-model/glm-4
+
+```ini
+# true or false
+DO_SAMPLE=
+```
+
 #### Chutes
 
 ```ini
@@ -558,6 +637,16 @@ API_KEY=
 MODEL=DeepSeek-V3-0324
 ```
 
+Only a few models support tool call as below:
+
+- Meta-Llama-3.1-8B-Instruct
+- Meta-Llama-3.1-405B-Instruct
+- Meta-Llama-3.3-70B-Instruct
+- Llama-4-Scout-17B-16E-Instruct
+- DeepSeek-V3-0324
+
+See official docs for more detail: https://docs.sambanova.ai/cloud/docs/capabilities/function-calling
+
 #### ModelScope
 
 ```ini
@@ -601,6 +690,45 @@ API_KEY=
 MODEL=command-a-03-2025
 ```
 
+Check official docs: https://docs.cohere.com/docs/text-gen-quickstart
+
+Support keys:
+
+```ini
+ENVIRONMENT=
+```
+
+For private deploy and Azure api, you need to set BASE_URL.
+
+```ini
+PROVIDER=cohere
+API_KEY=
+MODEL=command-a-03-2025
+BASE_URL=<YOUR_ENDPOINT>
+```
+
+For Bedrock and Sagemaker cohere api, you have to set below keys:
+
+See https://docs.cohere.com/docs/text-gen-quickstart.
+
+```ini
+PROVIDER=cohere-bedrock
+; PROVIDER=cohere-sagemaker
+API_KEY=
+MODEL=command-a-03-2025
+
+AWS_REGION=xx
+AWS_ACCESS_KEY_ID=xx
+AWS_SECRET_ACCESS_KEY=xx
+AWS_SESSION_TOKEN=xx
+```
+
+Note `MODEL` for Sagemaker should be endpoint name
+
+```ini
+MODEL=<YOUR_ENDPOINT_NAME>
+```
+
 #### Doubao
 
 You have to install doubao dependencies, `pip install 'yaicli[doubao]'`

{yaicli-0.6.2.dist-info → yaicli-0.6.4.dist-info}/RECORD

@@ -1,10 +1,10 @@
-pyproject.toml,sha256=xXPLxBBTEBHwusE6ko3ASXLhV3a5FV26eHNRAzkQzwc,2434
+pyproject.toml,sha256=LQv7NHuPZjn7h03OWDzftK8V0G_OG0626EkpVEUh4IA,2756
 yaicli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 yaicli/chat.py,sha256=_emvZEdgMBth2nQGaNWPf0P45oW2k3bpuIwqsxFcM5A,13676
-yaicli/cli.py,sha256=YUKbtvRcNbe6iTmm0dx-38QDzzrAP4cOYaOChxpJ2wg,23673
-yaicli/config.py,sha256=HrWYcelLXE61XX719eVcuuo3292xxf1BNQznWdvjQFQ,6535
+yaicli/cli.py,sha256=s8Bj4MSQmLblh4fHBPKS-DvJoAdMmp64KC7e7BwzmOs,23322
+yaicli/config.py,sha256=_xLHgyW5dDg76bL1SyTcrQCnVs4dFpXafIS3sClshK0,6563
 yaicli/console.py,sha256=vARPJd-3lafutsQWrGntQVjLrYqaJD3qisN82pmuhjU,1973
-yaicli/const.py,sha256=jC01jLLuuYJ7K_QaekXkIN7j1bIKCoGwJeTpquJy55Q,8178
+yaicli/const.py,sha256=G-EhMsSfOjKQLBhpOAc3pRtjvKmmWrwyyuyuGKHt7wk,8194
 yaicli/entry.py,sha256=Q1eqLE7tcHide7ooyPO7OCJpKE2YVuxR-NNFA2Pt2Hw,8693
 yaicli/exceptions.py,sha256=WBYg8OTJJzaj7lt6HE7ZyBoe5T6A3yZRNCRfWd4iN0c,372
 yaicli/history.py,sha256=s-57X9FMsaQHF7XySq1gGH_jpd_cHHTYafYu2ECuG6M,2472
@@ -16,29 +16,32 @@ yaicli/tools.py,sha256=xw8KEs_xlSf79A2Aq1rAsUWahS6A_e5QMLt7QDXL5bs,5086
 yaicli/utils.py,sha256=bpo3Xhozpxsaci3FtEIKZ32l4ZdyWMsrHjYGX0tB4J4,4541
 yaicli/functions/__init__.py,sha256=_FJooQ9GkijG8xLwuU0cr5GBrGnC9Nc6bnCeUjrsT0k,1271
 yaicli/functions/buildin/execute_shell_command.py,sha256=unl1-F8p6QZajeHdA0u5UpURMJM0WhdWMUWCCCHVRcI,1320
-yaicli/llms/__init__.py,sha256=cN54nu-YalZipXjLW0YAAe0rRv0tXAQ8lLi1ohuTpao,363
+yaicli/llms/__init__.py,sha256=x78cJujrJkelXPnzHS6pzHkITZdgLYZqJMnrMHbptoc,134
 yaicli/llms/client.py,sha256=mkE9KHSuPcJfpNQXbzF2YXGkel3jrOW8KfQ3YYpaK4M,4453
-yaicli/llms/provider.py,sha256=ey8PCq5IS6E9YleErVw_j6GtudtOawhFVBt3X8Be51U,3017
+yaicli/llms/provider.py,sha256=jF15kmY_tZVOjlw0fbHQkEvlmOX57-HBhILzG0KvXyo,3412
 yaicli/llms/providers/ai21_provider.py,sha256=SvgGj9_87KEqmxCMLbtsSkT8J3rUD7Mb21UF7pMWsks,3035
-yaicli/llms/providers/chatglm_provider.py,sha256=1xP4KVAi6SDKZ-lMi2wdzywtDydsTf6jDzh3jBBGMfA,6437
+yaicli/llms/providers/chatglm_provider.py,sha256=QEzALvY5FBhuDCx6rHjLi7GSRTwTHNUwo8gg4FEdrxs,6466
 yaicli/llms/providers/chutes_provider.py,sha256=mtvWvRRfHPH3JFfzym87wXtPNiMpLnur3805N9acx7E,882
-yaicli/llms/providers/cohere_provider.py,sha256=hc6vQxbCHz9kM2tNKK-kGkuOf4-gkskXW9ctr9V4Cxk,10837
-yaicli/llms/providers/deepseek_provider.py,sha256=VjGes_jFin5WGYNFxYKMoHwgAQX_eYbYhQKfjeh-9eI,438
+yaicli/llms/providers/cohere_provider.py,sha256=1UPzNqNOwM4_dsP4kvUaL9O6_bKjxm1lO6A0lM7hgS4,10959
+yaicli/llms/providers/deepseek_provider.py,sha256=iIV97x2ZCcwhGkshc8wpRi-YAnAnmo0n-YRegPlaOwQ,488
 yaicli/llms/providers/doubao_provider.py,sha256=4eOdE91ITUn3uo3mvYAzdrHsuFIIBwZWib21mtZn8OY,1938
+yaicli/llms/providers/gemini_provider.py,sha256=k_6JFmqiYPz5K8IioFic5tp8KAHgeeakjkPyqJVz8BI,8007
 yaicli/llms/providers/groq_provider.py,sha256=EiS1Yxw5jbAUBFCRYsJ57KYgZPk6oH-_gD72OfW8Oik,1358
-yaicli/llms/providers/infiniai_provider.py,sha256=1dseUIZiXsxYRATRtk_obFclyXMwi4glsP7l_tVtnv8,710
+yaicli/llms/providers/huggingface_provider.py,sha256=vDJyyK_aOlvktNvs-cji6pDtmKEp61vuVJ783BZw4pc,1247
+yaicli/llms/providers/infiniai_provider.py,sha256=8-nU6QE58PRoZL9b_HzbPp4yi6OGm7rXtfi9z7bJMOg,786
 yaicli/llms/providers/minimax_provider.py,sha256=W-j3dzrYMEv14bYt2pCPvPUxvxsUs-iMAcGB9yXakFs,744
-yaicli/llms/providers/modelscope_provider.py,sha256=BzBhYixiDEWB7gujQ0rcG__7nsv0psJRxdtYCYXBhdM,454
+yaicli/llms/providers/modelscope_provider.py,sha256=qWM0T7r0Zf8k3pLzjj7_IFdnmnx7S3rJO0f9rRm8-_A,504
 yaicli/llms/providers/ollama_provider.py,sha256=pjpYjfnHWnExweZi1KGbT07JGkcxzKPhqICo8dD82D0,6967
-yaicli/llms/providers/openai_provider.py,sha256=qEvzi4UHCQzwK4FQp1m2Ioahenp2uCQdFftTjUa6PvI,9739
+yaicli/llms/providers/openai_provider.py,sha256=ENn21QacP2iTcmbxuW7dgiw3_fUr8EGWhNSFR2yxjis,10079
 yaicli/llms/providers/openrouter_provider.py,sha256=R-7FrUrCAKPZ3gbnuo0M6rPlVw1mvSBjbLGs_FtZWM0,732
 yaicli/llms/providers/sambanova_provider.py,sha256=FFLrsvARt1UPAFWWgiuB6zvGzGKdtehKL58HdE1fo_M,2254
-yaicli/llms/providers/siliconflow_provider.py,sha256=7Ir73me9jGMO5TAZDjrAbX7tbb_QBmLjTGywY0yliqc,446
+yaicli/llms/providers/siliconflow_provider.py,sha256=CW2VSt6evUyFy21vN84Nvmw1P0JpmHBLznsgiXMnHM0,496
 yaicli/llms/providers/targon_provider.py,sha256=RQ808eS9lvsyvlzyKaQYcN0NimbpoNWgjHUzY1gLNs4,717
+yaicli/llms/providers/vertexai_provider.py,sha256=_ddrse1LfXRChTgkvxUlexyfJlfr0sVJH-Rmno3djSI,636
 yaicli/llms/providers/xai_provider.py,sha256=Q6iOvJZOXIAwRiiHMKEBgq8-W6SGVZ9QD1_532bNYfo,199
 yaicli/llms/providers/yi_provider.py,sha256=EnTm9qTxHPnzERsKqgGnzRIVhXFcAEdYqtOra65pGmY,719
-yaicli-0.6.2.dist-info/METADATA,sha256=rD4n886cYaYpdEyBugcrnJcKm6oiUIOMXLFczByLDiQ,53546
-yaicli-0.6.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-yaicli-0.6.2.dist-info/entry_points.txt,sha256=iYVyQP0PJIm9tQnlQheqT435kK_xdGoi5j9aswGV9hA,66
-yaicli-0.6.2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-yaicli-0.6.2.dist-info/RECORD,,
+yaicli-0.6.4.dist-info/METADATA,sha256=q1r4B-AADLIC9gAYEDd4BDHnAMnbjqDXrrOG3QNxbGc,55786
+yaicli-0.6.4.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+yaicli-0.6.4.dist-info/entry_points.txt,sha256=iYVyQP0PJIm9tQnlQheqT435kK_xdGoi5j9aswGV9hA,66
+yaicli-0.6.4.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+yaicli-0.6.4.dist-info/RECORD,,