Simple interactive shell added

README.md

@@ -86,6 +86,8 @@ cp examples/sgr_deep_research/config.yaml.example examples/sgr_deep_research/con
 
 ```bash
 sgr --config-file examples/sgr_deep_research/config.yaml
+# or use short option
+sgr -c examples/sgr_deep_research/config.yaml
 ```
 
 > **Note:** You can also run the server directly with Python:
@@ -94,6 +96,32 @@ sgr --config-file examples/sgr_deep_research/config.yaml
 > python -m sgr_agent_core.server --config-file examples/sgr_deep_research/config.yaml
 > ```
 
+### Using the CLI Tool (`sgrsh`)
+
+For interactive command-line usage, you can use the `sgrsh` utility:
+
+```bash
+# Single query mode
+sgrsh "Найди цену биткоина"
+
+# With agent selection (e.g. sgr_agent, dialog_agent)
+sgrsh --agent sgr_agent "What is AI?"
+
+# With custom config file
+sgrsh -c config.yaml -a sgr_agent "Your query"
+
+# Interactive chat mode (no query argument)
+sgrsh
+sgrsh -a sgr_agent
+```
+
+The `sgrsh` command:
+
+- Automatically looks for `config.yaml` in the current directory
+- Supports interactive chat mode for multiple queries
+- Handles clarification and dialog (intermediate results) requests from agents
+- Works with any agent defined in your configuration (e.g. `sgr_agent`, `dialog_agent`)
+
 For more examples and detailed usage instructions, see the [examples/](examples/) directory.
 
 ## Benchmarking

docs/en/getting-started/index.md

@@ -57,10 +57,61 @@ pip install sgr-agent-core
 
 See the [Installation Guide](installation.md) for detailed instructions and the [Using as Library](../framework/first-steps.md) guide to get started.
 
-### Next Steps
+### CLI Tool (`sgrsh`)
 
-- **[Using as Library](../framework/first-steps.md)** — Learn how to use SGR Agent Core as a Python library
-- **[API Server Quick Start](../sgr-api/SGR-Quick-Start.md)** — Get started with the REST API service
+After installation, you can use the `sgrsh` command-line tool for interactive agent usage:
+
+```bash
+# Single query mode
+sgrsh "Find the current Bitcoin price"
+
+# With agent selection
+sgrsh --agent sgr_agent "What is AI?"
+
+# With custom config file
+sgrsh -c config.yaml -a sgr_agent "Your query"
+
+# Interactive chat mode (no query argument)
+sgrsh
+sgrsh -a sgr_agent
+```
+
+The `sgrsh` command:
+- Automatically looks for `config.yaml` in the current directory
+- Supports interactive chat mode for multiple queries
+- Handles clarification requests from agents interactively
+- Works with any agent defined in your configuration
+
+### Using as Library
+
+```python
+import asyncio
+from sgr_agent_core import AgentDefinition, AgentFactory
+from sgr_agent_core.agents import SGRToolCallingAgent
+import sgr_agent_core.tools as tools
+
+async def main():
+    agent_def = AgentDefinition(
+        name="my_agent",
+        base_class=SGRToolCallingAgent,
+        tools=[tools.GeneratePlanTool, tools.FinalAnswerTool],
+        llm={
+            "api_key": "your-api-key",
+            "base_url": "https://api.openai.com/v1",
+        },
+    )
+
+    agent = await AgentFactory.create(
+        agent_def=agent_def,
+        task_messages=[{"role": "user", "content": "Research AI trends"}],
+    )
+
+    result = await agent.execute()
+    print(result)
+
+if __name__ == "__main__":
+    asyncio.run(main())
+```
 
 ## Documentation
 

docs/en/getting-started/installation.md

@@ -40,10 +40,17 @@ After installation, verify that the package is correctly installed:
 python -c "import sgr_agent_core; print(sgr_agent_core.__version__)"
 ```
 
-You should also be able to use the `sgr` command-line utility:
+You should also be able to use the command-line utilities:
 
 ```bash
+# API server command
 sgr --help
+# or with short option
+sgr -c config.yaml
+
+# Interactive CLI command
+sgrsh --help
+sgrsh "Your query here"
 ```
 
 ## Installation via Docker

docs/en/sgr-api/SGR-Summary-table.md

@@ -5,3 +5,4 @@ This table compares the available agent types in SGR Agent Core, showing their i
 | [SGRAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/sgr_agent.py) | `sgr_agent` | Structured Output  | ❌ Built into schema | 6 basic               | 1            | SO Union Type        |
 | [ToolCallingAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/tool_calling_agent.py) | `tool_calling_agent` | ❌ Absent          | ❌ Absent            | 6 basic               | 1            | FC "required"        |
 | [SGRToolCallingAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/sgr_tool_calling_agent.py) | `sgr_tool_calling_agent` | FC Tool enforced   | ✅ First step FC     | 7 (6 + ReasoningTool) | 2            | FC → FC    TOP AGENT |
+| [DialogAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/dialog_agent.py) | `dialog_agent` | Same as SGRToolCallingAgent | ✅ First step FC     | 8 (+ AnswerTool)       | 2            | FC → FC    Long dialogs |

docs/ru/getting-started/index.md

@@ -57,10 +57,61 @@ pip install sgr-agent-core
 
 См. [Руководство по установке](installation.md) для подробных инструкций и [Использование как библиотека](../framework/first-steps.md) для начала работы.
 
-### Следующие шаги
+### CLI утилита (`sgrsh`)
 
-- **[Использование как библиотека](../framework/first-steps.md)** — Узнайте, как использовать SGR Agent Core как Python библиотеку
-- **[Быстрый старт API сервера](../sgr-api/SGR-Quick-Start.md)** — Начните работу с REST API сервисом
+После установки вы можете использовать утилиту командной строки `sgrsh` для интерактивной работы с агентами:
+
+```bash
+# Режим одного запроса
+sgrsh "Найди текущую цену биткоина"
+
+# С выбором агента
+sgrsh --agent sgr_agent "Что такое AI?"
+
+# С указанием файла конфигурации
+sgrsh -c config.yaml -a sgr_agent "Ваш запрос"
+
+# Интерактивный режим чата (без аргумента запроса)
+sgrsh
+sgrsh -a sgr_agent
+```
+
+Команда `sgrsh`:
+- Автоматически ищет `config.yaml` в текущей директории
+- Поддерживает интерактивный режим чата для множественных запросов
+- Обрабатывает запросы на уточнение от агентов интерактивно
+- Работает с любым агентом, определённым в вашей конфигурации
+
+### Использование как библиотека
+
+```python
+import asyncio
+from sgr_agent_core import AgentDefinition, AgentFactory
+from sgr_agent_core.agents import SGRToolCallingAgent
+import sgr_agent_core.tools as tools
+
+async def main():
+    agent_def = AgentDefinition(
+        name="my_agent",
+        base_class=SGRToolCallingAgent,
+        tools=[tools.GeneratePlanTool, tools.FinalAnswerTool],
+        llm={
+            "api_key": "your-api-key",
+            "base_url": "https://api.openai.com/v1",
+        },
+    )
+
+    agent = await AgentFactory.create(
+        agent_def=agent_def,
+        task_messages=[{"role": "user", "content": "Исследуй тренды в AI"}],
+    )
+
+    result = await agent.execute()
+    print(result)
+
+if __name__ == "__main__":
+    asyncio.run(main())
+```
 
 ## Документация
 

docs/ru/getting-started/installation.md

@@ -40,10 +40,17 @@ pip install sgr-agent-core[docs]
 python -c "import sgr_agent_core; print(sgr_agent_core.__version__)"
 ```
 
-Также вы должны иметь возможность использовать утилиту командной строки `sgr`:
+Также вы должны иметь возможность использовать утилиты командной строки:
 
 ```bash
+# Команда API сервера
 sgr --help
+# или с коротким параметром
+sgr -c config.yaml
+
+# Интерактивная CLI команда
+sgrsh --help
+sgrsh "Ваш запрос здесь"
 ```
 
 ## Установка через Docker

docs/ru/sgr-api/SGR-Summary-table.md

@@ -5,3 +5,4 @@
 | [SGRAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/sgr_agent.py) | `sgr_agent` | Structured Output  | ❌ Встроен в схему | 6 базовых               | 1            | SO Union Type        |
 | [ToolCallingAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/tool_calling_agent.py) | `tool_calling_agent` | ❌ Отсутствует          | ❌ Отсутствует            | 6 базовых               | 1            | FC "required"        |
 | [SGRToolCallingAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/sgr_tool_calling_agent.py) | `sgr_tool_calling_agent` | FC Tool принудительно   | ✅ Первый шаг FC     | 7 (6 + ReasoningTool) | 2            | FC → FC    ЛУЧШИЙ АГЕНТ |
+| [DialogAgent](https://github.com/vamplabai/sgr-agent-core/blob/main/sgr_agent_core/agents/dialog_agent.py) | `dialog_agent` | Как SGRToolCallingAgent | ✅ Первый шаг FC     | 8 (+ AnswerTool)       | 2            | FC → FC    Длинные диалоги |

examples/sgr_deep_research/README.md

@@ -9,6 +9,7 @@ SGR Deep Research contains research agent definitions and configuration files fo
 - **SGR Agent** - Schema-Guided Reasoning agent for structured research
 - **Tool Calling Agent** - Function calling agent for research tasks
 - **SGR Tool Calling Agent** - Hybrid SGR + function calling agent
+- **Dialog Agent** - Dialog agent with intermediate results and long conversations
 
 All agents include:
 

examples/sgr_deep_research/config.yaml.example

@@ -5,7 +5,7 @@
 llm:
   api_key: "your-openai-api-key-here"  # Your OpenAI API key
   base_url: "https://api.openai.com/v1"  # API base URL
-  model: "gpt-4o-mini"  # Model name
+  model: "gpt-4.1-mini"  # Model name
   max_tokens: 8000  # Max output tokens
   temperature: 0.4  # Temperature (0.0-1.0)
   # proxy: "socks5://127.0.0.1:1081"  # Optional proxy (socks5:// or http://)
@@ -51,14 +51,16 @@ tools:
     # base_class defaults to sgr_agent_core.tools.AdaptPlanTool
   reasoning_tool:
     # base_class defaults to sgr_agent_core.tools.ReasoningTool
+  answer_tool:
+    # base_class defaults to sgr_agent_core.tools.AnswerTool
 
 # Agent Definitions
 agents:
   # SGR Agent for research
   sgr_agent:
     base_class: "agents.ResearchSGRAgent"
     llm:
-      model: "gpt-4o-mini"
+      model: "gpt-4.1-mini"
       temperature: 0.4
     tools:
       - "web_search_tool"
@@ -73,7 +75,7 @@ agents:
   tool_calling_agent:
     base_class: "agents.ResearchToolCallingAgent"
     llm:
-      model: "gpt-4o-mini"
+      model: "gpt-4.1-mini"
       temperature: 0.4
     tools:
       - "web_search_tool"
@@ -88,7 +90,7 @@ agents:
   sgr_tool_calling_agent:
     base_class: "agents.ResearchSGRToolCallingAgent"
     llm:
-      model: "gpt-4o-mini"
+      model: "gpt-4.1-mini"
       temperature: 0.4
     tools:
       - "web_search_tool"
@@ -99,3 +101,17 @@ agents:
       - "reasoning_tool"
       - "generate_plan_tool"
       - "adapt_plan_tool"
+
+  # Dialog Agent for research (intermediate results, long conversations)
+  dialog_agent:
+    base_class: "agents.ResearchDialogAgent"
+    llm:
+      model: "gpt-4.1-mini"
+      temperature: 0.4
+    tools:
+      - "web_search_tool"
+      - "extract_page_content_tool"
+      - "reasoning_tool"
+      - "answer_tool"
+      - "generate_plan_tool"
+      - "adapt_plan_tool"

examples/sgr_deep_research/definitions.py

@@ -8,6 +8,7 @@
 
 import sgr_agent_core.tools as tools
 from examples.sgr_deep_research.agents import (
+    ResearchDialogAgent,
     ResearchSGRAgent,
     ResearchSGRToolCallingAgent,
     ResearchToolCallingAgent,
@@ -50,5 +51,11 @@ def get_research_agents_definitions() -> dict[str, AgentDefinition]:
             tools=DEFAULT_TOOLKIT,
             prompts=PromptsConfig(system_prompt_file=Path("sgr_agent_core/prompts/research_system_prompt.txt")),
         ),
+        AgentDefinition(
+            name="research_dialog_agent",
+            base_class=ResearchDialogAgent,
+            tools=DEFAULT_TOOLKIT,
+            prompts=PromptsConfig(system_prompt_file=Path("sgr_agent_core/prompts/research_system_prompt.txt")),
+        ),
     ]
     return {agent.name: agent for agent in agents}

pyproject.toml

@@ -63,6 +63,7 @@ Documentation = "https://vamplabai.github.io/sgr-agent-core/"
 
 [project.scripts]
 sgr = "sgr_agent_core.server.__main__:main"
+sgrsh = "sgr_agent_core.cli.__main__:main"
 
 [project.optional-dependencies]
 dev = [

sgr_agent_core/agents/__init__.py

@@ -1,10 +1,12 @@
 """Agents module for SGR Agent Core."""
 
+from sgr_agent_core.agents.dialog_agent import DialogAgent
 from sgr_agent_core.agents.sgr_agent import SGRAgent
 from sgr_agent_core.agents.sgr_tool_calling_agent import SGRToolCallingAgent
 from sgr_agent_core.agents.tool_calling_agent import ToolCallingAgent
 
 __all__ = [
+    "DialogAgent",
     "SGRAgent",
     "SGRToolCallingAgent",
     "ToolCallingAgent",

sgr_agent_core/agents/dialog_agent.py

@@ -0,0 +1,52 @@
+"""Dialog agent for long-running conversations with intermediate results."""
+
+from typing import Type
+
+from openai import AsyncOpenAI
+
+from sgr_agent_core.agent_definition import AgentConfig
+from sgr_agent_core.agents.sgr_tool_calling_agent import SGRToolCallingAgent
+from sgr_agent_core.models import AgentStatesEnum
+from sgr_agent_core.tools import AnswerTool, BaseTool
+
+
+class DialogAgent(SGRToolCallingAgent):
+    """Agent specialized for dialog interactions with intermediate results.
+
+    Uses AnswerTool to share intermediate results and maintain
+    conversation flow, keeping the agent available for further
+    interactions. Supports long dialogs with full conversation history.
+    """
+
+    name: str = "dialog_agent"
+
+    def __init__(
+        self,
+        task_messages: list,
+        openai_client: AsyncOpenAI,
+        agent_config: AgentConfig,
+        toolkit: list[Type[BaseTool]],
+        def_name: str | None = None,
+        **kwargs: dict,
+    ):
+        # Ensure AnswerTool is in toolkit for dialog flow; keep tools from config/registry
+        answer_toolkit = [AnswerTool]
+        merged_toolkit = answer_toolkit + [t for t in toolkit if t is not AnswerTool]
+        super().__init__(
+            task_messages=task_messages,
+            openai_client=openai_client,
+            agent_config=agent_config,
+            toolkit=merged_toolkit,
+            def_name=def_name,
+            **kwargs,
+        )
+
+    async def _after_action_phase(self, action_tool: BaseTool, result: str) -> None:
+        """Wait for user response when AnswerTool was used."""
+        await super()._after_action_phase(action_tool, result)
+        if isinstance(action_tool, AnswerTool):
+            self.logger.info("\n💬 Dialog shared - agent waiting for response")
+            self._context.state = AgentStatesEnum.WAITING_FOR_CLARIFICATION
+            self.streaming_generator.finish(result)
+            self._context.clarification_received.clear()
+            await self._context.clarification_received.wait()