feat: agent&tool docs/tests/examples (#319)

* feat: init agent&tool architecture

* feat: agent&tool docs/tests/examples

* fix bugs

Author: seancoding-day

.github/workflows/IntegrationTest.yml

@@ -26,6 +26,7 @@ jobs:
         pip install pytest
         if [ -f requirements/runtime.txt ]; then pip install -r requirements/runtime.txt; fi
         pip install pyspark
+        pip install tavily-python
         pip install -e .
 
     - name: Check Python syntax and imports

README.md

@@ -69,7 +69,7 @@
 
 🤖 **RAG System Assessment** - Comprehensive evaluation of retrieval and generation quality with 5 academic-backed metrics
 
-🧠 **LLM & Rule Hybrid** - Combine fast heuristic rules (30+ built-in) with LLM-based deep assessment
+🧠 **LLM & Rule & Agent Hybrid** - Combine fast heuristic rules (30+ built-in) with LLM-based deep assessment
 
 🚀 **Flexible Execution** - Run locally for rapid iteration or scale with Spark for billion-scale datasets
 
@@ -384,6 +384,12 @@ input_data = {
 ✅ Vision-Language Models (InternVL, Gemini)  
 ✅ Custom prompt registration
 
+**Agent-Based** - Multi-step reasoning with tools
+✅ Web search integration (Tavily)
+✅ Adaptive context gathering
+✅ Multi-source fact verification
+✅ Custom agent & tool registration
+
 **Extensible Architecture**  
 ✅ Plugin-based rule/prompt/model registration  
 ✅ Clean separation of concerns (agents, tools, orchestration)  
@@ -492,6 +498,65 @@ class CustomEvaluator(BaseOpenAI):
 - [Custom Rules](examples/register/sdk_register_rule.py)
 - [Custom Models](examples/register/sdk_register_llm.py)
 
+### Agent-Based Evaluation with Tools
+
+Dingo supports agent-based evaluators that can use external tools for multi-step reasoning and adaptive context gathering:
+
+```python
+from dingo.io import Data
+from dingo.io.output.eval_detail import EvalDetail
+from dingo.model import Model
+from dingo.model.llm.agent.base_agent import BaseAgent
+
+@Model.llm_register('MyAgent')
+class MyAgent(BaseAgent):
+    """Custom agent with tool support"""
+
+    available_tools = ["tavily_search", "my_custom_tool"]
+    max_iterations = 5
+
+    @classmethod
+    def eval(cls, input_data: Data) -> EvalDetail:
+        # Use tools for fact-checking
+        search_result = cls.execute_tool('tavily_search', query=input_data.content)
+
+        # Multi-step reasoning with LLM
+        result = cls.send_messages([...])
+
+        return EvalDetail(...)
+```
+
+**Built-in Agent:**
+- `AgentHallucination`: Enhanced hallucination detection with web search fallback
+
+**Configuration Example:**
+```json
+{
+  "evaluator": [{
+    "evals": [{
+      "name": "AgentHallucination",
+      "config": {
+        "key": "openai-api-key",
+        "model": "gpt-4",
+        "parameters": {
+          "agent_config": {
+            "max_iterations": 5,
+            "tools": {
+              "tavily_search": {"api_key": "tavily-key"}
+            }
+          }
+        }
+      }
+    }]
+  }]
+}
+```
+
+**Learn More:**
+- [Agent Development Guide](docs/agent_development_guide.md) - Comprehensive guide for creating custom agents and tools
+- [AgentHallucination Example](examples/agent/agent_hallucination_example.py) - Production agent example
+- [AgentFactCheck Example](examples/agent/agent_executor_example.py) - LangChain agent example
+
 ## ⚙️ Execution Modes
 
 ### Local Executor (Development & Small-Scale)

README_zh-CN.md

@@ -68,7 +68,7 @@
 
 🤖 **RAG 系统评估** - 使用 5 个学术支持的指标全面评估检索和生成质量
 
-🧠 **LLM 与规则混合** - 结合快速启发式规则（30+ 内置规则）和基于 LLM 的深度评估
+🧠 **LLM、规则和智能体混合** - 结合快速启发式规则（30+ 内置规则）和基于 LLM 的深度评估
 
 🚀 **灵活执行** - 本地运行快速迭代，或使用 Spark 扩展到数十亿级数据集
 
@@ -383,6 +383,12 @@ input_data = {
 ✅ 视觉语言模型（InternVL、Gemini）  
 ✅ 自定义 prompt 注册
 
+**基于智能体** - 多步推理与工具
+✅ 网络搜索集成（Tavily）
+✅ 自适应上下文收集
+✅ 多源事实验证
+✅ 自定义智能体与工具注册
+
 **可扩展架构**  
 ✅ 基于插件的规则/prompt/模型注册  
 ✅ 清晰的关注点分离（agents、tools、orchestration）  
@@ -486,6 +492,65 @@ class MyCustomModel(BaseOpenAI):
 - [注册规则](examples/register/sdk_register_rule.py)
 - [注册模型](examples/register/sdk_register_llm.py)
 
+### 智能体评估与工具
+
+Dingo 支持基于智能体的评估器，可以使用外部工具进行多步推理和自适应上下文收集：
+
+```python
+from dingo.io import Data
+from dingo.io.output.eval_detail import EvalDetail
+from dingo.model import Model
+from dingo.model.llm.agent.base_agent import BaseAgent
+
+@Model.llm_register('MyAgent')
+class MyAgent(BaseAgent):
+    """支持工具的自定义智能体"""
+
+    available_tools = ["tavily_search", "my_custom_tool"]
+    max_iterations = 5
+
+    @classmethod
+    def eval(cls, input_data: Data) -> EvalDetail:
+        # 使用工具进行事实核查
+        search_result = cls.execute_tool('tavily_search', query=input_data.content)
+
+        # 使用LLM进行多步推理
+        result = cls.send_messages([...])
+
+        return EvalDetail(...)
+```
+
+**内置智能体：**
+- `AgentHallucination`: 增强的幻觉检测，支持网络搜索回退
+
+**配置示例：**
+```json
+{
+  "evaluator": [{
+    "evals": [{
+      "name": "AgentHallucination",
+      "config": {
+        "key": "openai-api-key",
+        "model": "gpt-4",
+        "parameters": {
+          "agent_config": {
+            "max_iterations": 5,
+            "tools": {
+              "tavily_search": {"api_key": "tavily-key"}
+            }
+          }
+        }
+      }
+    }]
+  }]
+}
+```
+
+**了解更多：**
+- [智能体开发指南](docs/agent_development_guide.md)
+- [AgentHallucination 示例](examples/agent/agent_hallucination_example.py)
+- [AgentFactCheck LangChain示例](examples/agent/agent_executor_example.py)
+
 ## 执行引擎
 
 ### 本地执行