update wendong

2026-03-22 05:57:17 +08:00 · 2025-03-15 12:24:31 +08:00
parent c7c94a233a
commit 05aebff028
18 changed files with 1398 additions and 115 deletions
--- a/README.md
+++ b/README.md
@@ -224,7 +224,7 @@ OWL requires various API keys to interact with different services. The `owl/.env
 2. **Configure Your API Keys**:
   Open the `.env` file in your preferred text editor and insert your API keys in the corresponding fields.
   
-   > **Note**: For the minimal example (`run_mini.py`), you only need to configure the LLM API key (e.g., `OPENAI_API_KEY`).
+   > **Note**: For the minimal example (`examples/run_mini.py`), you only need to configure the LLM API key (e.g., `OPENAI_API_KEY`).

 ### Option 2: Setting Environment Variables Directly

@@ -275,7 +275,7 @@ cd .. && source .venv/bin/activate && cd owl
 playwright install-deps 

 #run example demo script
-xvfb-python run.py
+xvfb-python examples/run.py

 # Option 2: Build and run using the provided scripts
 cd .container
@@ -299,17 +299,17 @@ npx -y @smithery/cli install @wonderwhy-er/desktop-commander --client claude
 npx @wonderwhy-er/desktop-commander setup

 # Run the MCP example
-python owl/run_mcp.py
+python examples/run_mcp.py
 ```

-This example showcases how OWL agents can seamlessly interact with file systems, web automation, and information retrieval through the MCP protocol. Check out `owl/run_mcp.py` for the full implementation.
+This example showcases how OWL agents can seamlessly interact with file systems, web automation, and information retrieval through the MCP protocol. Check out `examples/run_mcp.py` for the full implementation.

 ## Basic Usage

 After installation and setting up your environment variables, you can start using OWL right away:

 ```bash
-python owl/run.py
+python examples/run.py
 ```

 ## Running with Different Models
@@ -330,28 +330,28 @@ OWL supports various LLM backends, though capabilities may vary depending on the

 ```bash
 # Run with Qwen model
-python owl/examples/run_qwen_zh.py
+python examples/run_qwen_zh.py

 # Run with Deepseek model
-python owl/examples/run_deepseek_zh.py
+python examples/run_deepseek_zh.py

 # Run with other OpenAI-compatible models
-python owl/examples/run_openai_compatiable_model.py
+python examples/run_openai_compatiable_model.py

 # Run with Azure OpenAI
-python owl/run_azure_openai.py
+python examples/run_azure_openai.py

 # Run with Ollama
-python owl/examples/run_ollama.py
+python examples/run_ollama.py
 ```

 For a simpler version that only requires an LLM API key, you can try our minimal example:

 ```bash
-python owl/examples/run_mini.py
+python examples/run_mini.py
 ```

-You can run OWL agent with your own task by modifying the `run.py` script:
+You can run OWL agent with your own task by modifying the `examples/run.py` script:

 ```python
 # Define your own task
@@ -393,7 +393,7 @@ Here are some tasks you can try with OWL:

 OWL's MCP integration provides a standardized way for AI models to interact with various tools and data sources:

-Try our comprehensive MCP example in `owl/run_mcp.py` to see these capabilities in action!
+Try our comprehensive MCP example in `examples/run_mcp.py` to see these capabilities in action!

 ## Available Toolkits

@@ -464,10 +464,10 @@ OWL includes an intuitive web-based user interface that makes it easier to inter

 ```bash
 # Start the Chinese version
-python owl/webapp_zh.py
+python examples/webapp_zh.py

 # Start the English version
-python owl/webapp.py
+python examples/webapp.py
 ```

 ## Features
@@ -545,7 +545,7 @@ Join us ([*Discord*](https://discord.camel-ai.org/) or [*WeChat*](https://ghli.o

 Join us for further discussions!
 <!-- ![](./assets/community.png) -->
-![](./assets/community_8.jpg)
+![](./assets/community.jpg)

 # ❓ FAQ

--- a/README_zh.md
+++ b/README_zh.md
@@ -219,7 +219,7 @@ OWL 需要各种 API 密钥来与不同的服务进行交互。`owl/.env_templat
 2. **配置你的 API 密钥**：
   在你喜欢的文本编辑器中打开 `.env` 文件，并在相应字段中插入你的 API 密钥。
   
-   > **注意**：对于最小示例（`run_mini.py`），你只需要配置 LLM API 密钥（例如，`OPENAI_API_KEY`）。
+   > **注意**：对于最小示例（`examples/run_mini.py`），你只需要配置 LLM API 密钥（例如，`OPENAI_API_KEY`）。

 ### 选项 2：直接设置环境变量

@@ -269,7 +269,7 @@ cd .. && source .venv/bin/activate && cd owl
 playwright install-deps 

 #运行例子演示脚本
-xvfb-python run.py
+xvfb-python examples/run.py

 # 选项2：使用提供的脚本构建和运行
 cd .container
@@ -293,23 +293,23 @@ npx -y @smithery/cli install @wonderwhy-er/desktop-commander --client claude
 npx @wonderwhy-er/desktop-commander setup

 # 运行 MCP 示例
-python owl/run_mcp.py
+python examples/run_mcp.py
 ```

-这个示例展示了 OWL 智能体如何通过 MCP 协议无缝地与文件系统、网页自动化和信息检索进行交互。查看 `owl/run_mcp.py` 了解完整实现。
+这个示例展示了 OWL 智能体如何通过 MCP 协议无缝地与文件系统、网页自动化和信息检索进行交互。查看 `examples/run_mcp.py` 了解完整实现。

 ## 基本用法
   
 运行以下示例：

 ```bash
-python owl/run.py
+python examples/run.py
 ```

 我们还提供了一个最小化示例，只需配置LLM的API密钥即可运行：

 ```bash
-python owl/run_mini.py
+python examples/run_mini.py
 ```

 ## 使用不同的模型
@@ -330,22 +330,22 @@ OWL 支持多种 LLM 后端，但功能可能因模型的工具调用和多模

 ```bash
 # 使用 Qwen 模型运行
-python owl/examples/run_qwen_zh.py
+python examples/run_qwen_zh.py

 # 使用 Deepseek 模型运行
-python owl/examples/run_deepseek_zh.py
+python examples/run_deepseek_zh.py

 # 使用其他 OpenAI 兼容模型运行
-python owl/examples/run_openai_compatiable_model.py
+python examples/run_openai_compatiable_model.py

 # 使用 Azure OpenAI模型运行
-python owl/run_azure_openai.py
+python examples/run_azure_openai.py

 # 使用 Ollama 运行
-python owl/examples/run_ollama.py
+python examples/run_ollama.py
 ```

-你可以通过修改 `run.py` 脚本来运行自己的任务：
+你可以通过修改 `examples/run.py` 脚本来运行自己的任务：

 ```python
 # Define your own task
@@ -383,7 +383,7 @@ OWL 将自动调用与文档相关的工具来处理文件并提取答案。

 OWL 的 MCP 集成为 AI 模型与各种工具和数据源的交互提供了标准化的方式。

-查看我们的综合示例 `owl/run_mcp.py` 来体验这些功能！
+查看我们的综合示例 `examples/run_mcp.py` 来体验这些功能！

 ## 可用工具包

@@ -479,7 +479,7 @@ git checkout gaia58.18

 2. 运行评估脚本：
 ```bash
-python run_gaia_roleplaying.py
+python examples/run_gaia_roleplaying.py
 ```

 # ⏱️ 未来计划
@@ -531,7 +531,7 @@ python run_gaia_roleplaying.py

 加入我们，参与更多讨论！
 <!-- ![](./assets/community.png) -->
-![](./assets/community_8.jpg)
+![](./assets/community.jpg)
 <!-- ![](./assets/meetup.jpg) -->

 # ❓ 常见问题
--- a/owl/examples/run.py
+++ b/owl/examples/run.py
--- a/examples/run_azure_openai.py
+++ b/examples/run_azure_openai.py
--- a/owl/examples/run_deepseek_zh.py
+++ b/owl/examples/run_deepseek_zh.py
--- a/owl/examples/run_gaia_roleplaying.py
+++ b/owl/examples/run_gaia_roleplaying.py
--- a/examples/run_mcp.py
+++ b/examples/run_mcp.py
--- a/owl/examples/run_mini.py
+++ b/owl/examples/run_mini.py
--- a/owl/examples/run_ollama.py
+++ b/owl/examples/run_ollama.py
--- a/owl/examples/run_openai_compatiable_model.py
+++ b/owl/examples/run_openai_compatiable_model.py
--- a/owl/examples/run_qwen_mini_zh.py
+++ b/owl/examples/run_qwen_mini_zh.py
--- a/owl/examples/run_qwen_zh.py
+++ b/owl/examples/run_qwen_zh.py
--- a/owl/examples/run_terminal.py
+++ b/owl/examples/run_terminal.py
--- a/owl/examples/run_terminal_zh.py
+++ b/owl/examples/run_terminal_zh.py
@@ -25,7 +25,6 @@ from camel.logger import set_log_level

 from owl.utils import run_society
 from camel.societies import RolePlaying
-import os

 load_dotenv()
 set_log_level(level="DEBUG")
--- a/owl/.env_template
+++ b/owl/.env_template
@@ -4,7 +4,7 @@
 #===========================================

 # OPENAI API (https://platform.openai.com/api-keys)
-# OPENAI_API_KEY= ""
+OPENAI_API_KEY='Your_Key'
 # OPENAI_API_BASE_URL=""

 # Azure OpenAI API
@@ -15,22 +15,22 @@


 # Qwen API (https://help.aliyun.com/zh/model-studio/developer-reference/get-api-key)
-# QWEN_API_KEY=""
+QWEN_API_KEY='Your_Key'

 # DeepSeek API (https://platform.deepseek.com/api_keys)
-# DEEPSEEK_API_KEY=""
+DEEPSEEK_API_KEY='Your_Key'

 #===========================================
 # Tools & Services API
 #===========================================

-# Google Search API (https://developers.google.com/custom-search/v1/overview)
-# GOOGLE_API_KEY=""
-# SEARCH_ENGINE_ID=""
+# Google Search API (https://coda.io/@jon-dallas/google-image-search-pack-example/search-engine-id-and-google-api-key-3)
+GOOGLE_API_KEY='Your_Key'
+SEARCH_ENGINE_ID='Your_ID'

 # Chunkr API (https://chunkr.ai/)
-# CHUNKR_API_KEY=""
+CHUNKR_API_KEY='Your_Key'

 # Firecrawl API (https://www.firecrawl.dev/)
-#FIRECRAWL_API_KEY=""
+FIRECRAWL_API_KEY='Your_Key'
 #FIRECRAWL_API_URL="https://api.firecrawl.dev"
--- a/owl/utils/enhanced_role_playing.py
+++ b/owl/utils/enhanced_role_playing.py
@@ -461,6 +461,10 @@ def run_society(
            assistant_response.info["usage"]["completion_tokens"]
            + user_response.info["usage"]["completion_tokens"]
        )
+        overall_prompt_token_count += (
+            assistant_response.info["usage"]["prompt_tokens"]
+            + user_response.info["usage"]["prompt_tokens"]
+        )

        # convert tool call to dict
        tool_call_records: List[dict] = []
--- a/owl/webapp.py
+++ b/owl/webapp.py
--- a/owl/webapp_zh.py
+++ b/owl/webapp_zh.py
@@ -151,7 +151,7 @@ def get_latest_logs(max_lines=100, queue_source=None):

    # 如果仍然没有日志，返回提示信息
    if not logs:
-        return "暂无对话记录。"
+        return "初始化运行中..."

    # 过滤日志，只保留 camel.agents.chat_agent - INFO 的日志
    filtered_logs = []
@@ -242,87 +242,49 @@ MODULE_DESCRIPTIONS = {
    "run": "默认模式：使用OpenAI模型的默认的智能体协作模式，适合大多数任务。",
    "run_mini": "使用使用OpenAI模型最小化配置处理任务",
    "run_deepseek_zh": "使用deepseek模型处理中文任务",
-    "run_terminal_zh": "终端模式：可执行命令行操作，支持网络搜索、文件处理等功能。适合需要系统交互的任务，使用OpenAI模型",
-    "run_gaia_roleplaying": "GAIA基准测试实现，用于评估Agent能力",
    "run_openai_compatiable_model": "使用openai兼容模型处理任务",
    "run_ollama": "使用本地ollama模型处理任务",
    "run_qwen_mini_zh": "使用qwen模型最小化配置处理任务",
    "run_qwen_zh": "使用qwen模型处理任务",
 }

-# API帮助信息
-API_HELP_INFO = {
-    "OPENAI_API_KEY": {
-        "name": "OpenAI API",
-        "desc": "OpenAI API密钥，用于访问GPT系列模型",
-        "url": "https://platform.openai.com/api-keys",
-    },
-    "QWEN_API_KEY": {
-        "name": "通义千问 API",
-        "desc": "阿里云通义千问API密钥",
-        "url": "https://help.aliyun.com/zh/model-studio/developer-reference/get-api-key",
-    },
-    "DEEPSEEK_API_KEY": {
-        "name": "DeepSeek API",
-        "desc": "DeepSeek API密钥",
-        "url": "https://platform.deepseek.com/api_keys",
-    },
-    "GOOGLE_API_KEY": {
-        "name": "Google Search API",
-        "desc": "Google自定义搜索API密钥",
-        "url": "https://developers.google.com/custom-search/v1/overview",
-    },
-    "SEARCH_ENGINE_ID": {
-        "name": "Google Search Engine ID",
-        "desc": "Google自定义搜索引擎ID",
-        "url": "https://developers.google.com/custom-search/v1/overview",
-    },
-    "HF_TOKEN": {
-        "name": "Hugging Face API",
-        "desc": "Hugging Face API令牌",
-        "url": "https://huggingface.co/join",
-    },
-    "CHUNKR_API_KEY": {
-        "name": "Chunkr API",
-        "desc": "Chunkr API密钥",
-        "url": "https://chunkr.ai/",
-    },
-    "FIRECRAWL_API_KEY": {
-        "name": "Firecrawl API",
-        "desc": "Firecrawl API密钥",
-        "url": "https://www.firecrawl.dev/",
-    },
-}

 # 默认环境变量模板
-DEFAULT_ENV_TEMPLATE = """# MODEL & API (See https://docs.camel-ai.org/key_modules/models.html#)
+DEFAULT_ENV_TEMPLATE = """#===========================================
+# MODEL & API 
+# (See https://docs.camel-ai.org/key_modules/models.html#)
+#===========================================

-# OPENAI API
-# OPENAI_API_KEY= ""
+# OPENAI API (https://platform.openai.com/api-keys)
+OPENAI_API_KEY='Your_Key'
 # OPENAI_API_BASE_URL=""

+# Azure OpenAI API
+# AZURE_OPENAI_BASE_URL=""
+# AZURE_API_VERSION=""
+# AZURE_OPENAI_API_KEY=""
+# AZURE_DEPLOYMENT_NAME=""
+
+
 # Qwen API (https://help.aliyun.com/zh/model-studio/developer-reference/get-api-key)
-# QWEN_API_KEY=""
+QWEN_API_KEY='Your_Key'

 # DeepSeek API (https://platform.deepseek.com/api_keys)
-# DEEPSEEK_API_KEY=""
+DEEPSEEK_API_KEY='Your_Key'

 #===========================================
 # Tools & Services API
 #===========================================

-# Google Search API (https://developers.google.com/custom-search/v1/overview)
-GOOGLE_API_KEY=""
-SEARCH_ENGINE_ID=""
-
-# Hugging Face API (https://huggingface.co/join)
-HF_TOKEN=""
+# Google Search API (https://coda.io/@jon-dallas/google-image-search-pack-example/search-engine-id-and-google-api-key-3)
+GOOGLE_API_KEY='Your_Key'
+SEARCH_ENGINE_ID='Your_ID'

 # Chunkr API (https://chunkr.ai/)
-CHUNKR_API_KEY=""
+CHUNKR_API_KEY='Your_Key'

 # Firecrawl API (https://www.firecrawl.dev/)
-FIRECRAWL_API_KEY=""
+FIRECRAWL_API_KEY='Your_Key'
 #FIRECRAWL_API_URL="https://api.firecrawl.dev"
 """

@@ -357,7 +319,7 @@ def run_owl(question: str, example_module: str) -> Tuple[str, str, str]:
    # 验证输入
    if not validate_input(question):
        logging.warning("用户提交了无效的输入")
-        return ("请输入有效的问题", "0", "❌ 错误: 输入无效")
+        return ("请输入有效的问题", "0", "❌ 错误: 输入问题无效")

    try:
        # 确保环境变量已加载
@@ -374,7 +336,7 @@ def run_owl(question: str, example_module: str) -> Tuple[str, str, str]:
            )

        # 动态导入目标模块
-        module_path = f"owl.examples.{example_module}"
+        module_path = f"examples.{example_module}"
        try:
            logging.info(f"正在导入模块: {module_path}")
            module = importlib.import_module(module_path)
@@ -452,8 +414,6 @@ def update_module_description(module_name: str) -> str:
    return MODULE_DESCRIPTIONS.get(module_name, "无可用描述")


-# 环境变量管理功能
-
 # 存储前端配置的环境变量
 WEB_FRONTEND_ENV_VARS: dict[str, str] = {}

@@ -646,7 +606,9 @@ def get_api_guide(key: str) -> str:
    elif "deepseek" in key_lower:
        return "https://platform.deepseek.com/api_keys"
    elif "google" in key_lower:
-        return "https://developers.google.com/custom-search/v1/overview"
+        return "https://coda.io/@jon-dallas/google-image-search-pack-example/search-engine-id-and-google-api-key-3"
+    elif "search_engine_id" in key_lower:
+        return "https://coda.io/@jon-dallas/google-image-search-pack-example/search-engine-id-and-google-api-key-3"
    elif "chunkr" in key_lower:
        return "https://chunkr.ai/"
    elif "firecrawl" in key_lower:
@@ -701,11 +663,11 @@ def save_env_table_changes(data):

            # 遍历DataFrame的每一行
            for index, row in data.iterrows():
-                # 使用列名或索引访问数据
+                # 使用列名访问数据
                if len(columns) >= 3:
-                    # 如果有列名，使用列名访问
-                    key = row.iloc[1] if hasattr(row, "iloc") else row[1]
-                    value = row.iloc[2] if hasattr(row, "iloc") else row[2]
+                    # 获取变量名和值 (第0列是变量名，第1列是值)
+                    key = row[0] if isinstance(row, pd.Series) else row.iloc[0]
+                    value = row[1] if isinstance(row, pd.Series) else row.iloc[1]

                    # 检查是否为空行或已删除的变量
                    if key and str(key).strip():  # 如果键名不为空，则添加或更新
@@ -812,6 +774,9 @@ def create_ui():
        """处理问题并实时更新日志"""
        global CURRENT_PROCESS

+        # 清空日志文件
+        clear_log_file()
+
        # 创建一个后台线程来处理问题
        result_queue = queue.Queue()

@@ -874,6 +839,8 @@ def create_ui():
                # 🦉 OWL 多智能体协作系统

                基于CAMEL框架开发的先进多智能体协作系统，旨在通过智能体协作解决复杂问题。
+                可以通过修改本地脚本自定义模型和工具。
+                本网页应用目前处于测试阶段，仅供演示和测试使用，尚未推荐用于生产环境。
                """
        )

@@ -1082,6 +1049,7 @@ def create_ui():
                    label="问题",
                    elem_id="question_input",
                    show_copy_button=True,
+                    value="打开百度搜索，总结一下camel-ai的camel框架的github star、fork数目等，并把数字用plot包写成python文件保存到本地，并运行生成的python文件。",
                )

                # 增强版模块选择下拉菜单
@@ -1141,7 +1109,7 @@ def create_ui():
                        gr.Markdown("""
                            ## 环境变量管理
                            
-                            在此处设置模型API密钥和其他服务凭证。这些信息将保存在本地的`.env`文件中，确保您的API密钥安全存储且不会上传到网络。
+                            在此处设置模型API密钥和其他服务凭证。这些信息将保存在本地的`.env`文件中，确保您的API密钥安全存储且不会上传到网络。正确设置API密钥对于OWL系统的功能至关重要, 可以按找工具需求灵活配置环境变量。
                            """)

                        # 主要内容分为两列布局
@@ -1150,12 +1118,9 @@ def create_ui():
                            with gr.Column(scale=3):
                                with gr.Box(elem_classes="env-controls"):
                                    # 环境变量表格 - 设置为可交互以直接编辑
-                                    gr.Markdown("### 环境变量管理")
                                    gr.Markdown("""
-                                    管理您的API密钥和其他环境变量。正确设置API密钥对于OWL系统的功能至关重要。
-                                    
                                    <div style="background-color: #e7f3fe; border-left: 6px solid #2196F3; padding: 10px; margin: 15px 0; border-radius: 4px;">
-                                      <strong>提示：</strong> 请确保正确设置API密钥以确保系统功能正常
+                                      <strong>提示：</strong> 请确保运行cp .env_template .env创建本地.env文件，根据运行模块灵活配置所需环境变量
                                    </div>
                                    """)

@@ -1186,7 +1151,6 @@ def create_ui():
                                      <li><strong>删除变量</strong>: 清空变量名即可删除该行</li>
                                      <li><strong>获取API密钥</strong>: 点击"获取指南"列中的链接获取相应API密钥</li>
                                    </ul>
-                                    <strong>注意</strong>: 所有API密钥都安全地存储在本地，不会上传到网络
                                    </div>
                                    """,
                                        elem_classes="env-instructions",
@@ -1221,7 +1185,7 @@ def create_ui():

        # 示例问题
        examples = [
-            "打开百度搜索，总结一下camel-ai的camel框架的github star、fork数目等，并把数字用plot包写成python文件保存到本地，用本地终端执行python文件显示图出来给我",
+            "打开百度搜索，总结一下camel-ai的camel框架的github star、fork数目等，并把数字用plot包写成python文件保存到本地，并运行生成的python文件。",
            "浏览亚马逊并找出一款对程序员有吸引力的产品。请提供产品名称和价格",
            "写一个hello world的python文件，保存到本地",
        ]