๐ซ CAMEL: The first and the best multi-agent framework. Finding the Scaling Law of Agents. https://www.camel-ai.org
Python agentai-societiesartificial-intelligencecommunicative-aicooperative-aideep-learninglarge-language-modelsmulti-agent-systemsnatural-language-processing
usecase: Research assistant (#2484)
Bipul707015491604c0f07
diff --git a/examples/usecases/multi_agent_research_assistant/.env.template b/examples/usecases/multi_agent_research_assistant/.env.templatenew file mode 100644index 000000000..b71a0d619--- /dev/null+++ b/examples/usecases/multi_agent_research_assistant/.env.template@@ -0,0 +1,8 @@+GOOGLE_API_KEY=''+SEARCH_ENGINE_ID=''+OPENAI_API_KEY=''+DEEPSEEK_API_KEY=''+CHUNKR_API_KEY=''+FIRECRAWL_API_KEY=''+ASKNEWS_CLIENT_ID=''+ASKNEWS_CLIENT_SECRET=''diff --git a/examples/usecases/multi_agent_research_assistant/README.md b/examples/usecases/multi_agent_research_assistant/README.mdnew file mode 100644index 000000000..71209cd08--- /dev/null+++ b/examples/usecases/multi_agent_research_assistant/README.md@@ -0,0 +1,85 @@+# ๐ง CAMEL Multi-Agent Research Assistant++A powerful, AI-driven research assistant built using the [CAMEL-AI](https://www.camel-ai.org/) framework. This application leverages multiple specialized toolkits and a role-playing agent architecture to autonomously generate comprehensive research reports on any given topic.++## ๐ Features++- **Multi-Agent Collaboration**: Utilizes CAMEL's role-playing framework to simulate interactions between a "Project Manager" and a "Researcher Agent" for dynamic task execution.+- **Diverse Toolkit Integration**: Incorporates various CAMEL toolkits including:+ - `GoogleScholarToolkit` and `SemanticScholarToolkit` for academic paper retrieval.+ - `ArxivToolkit` for accessing preprints.+ - `AskNewsToolkit` for fetching relevant news articles.+ - `ThinkingToolkit` for planning and synthesis.+ - `FileWriteToolkit` for saving reports locally.+ - `LinkedInToolkit` for potential dissemination.+ - `DalleToolkit` for generating illustrative images.+- **Streamlit Interface**: Provides an intuitive web interface for users to input topics and receive generated reports.++## ๐ ๏ธ Installation++1. **Clone the Repository**:++ ```bash+ git clone https://github.com/camel-ai/camel.git+ cd camel/examples/usecases/multi_agent_research_assistant+ ```++2. **Create and Activate a Virtual Environment**:++ ```bash+ python -m venv venv+ source venv/bin/activate # On Windows: venv\Scripts\activate+ ```++3. **Install Dependencies**:++ ```bash+ pip install -r requirements.txt+ ```++4. ## ๐ ๏ธ Setup Environment Variables++ Copy the provided `.env.template` to `.env`:++ Open the `.env` file in your preferred editor and add your required API keys. At a minimum, you need to provide your OpenAI API key:++ ```env+ OPENAI_API_KEY=your_openai_api_key+ ```++ Replace `your_openai_api_key` with your actual key from OpenAI.++ Additional environment variables may be required depending on the toolkits you enable (e.g., for LinkedIn, DALLยทE, news APIs, etc.).+++## โถ๏ธ Running the Application++Start the Streamlit application:++```bash+streamlit run app.py+```++This will open the application in your default web browser. Enter a research topic and click "Generate Report" to initiate the multi-agent research process.++## ๐ Project Structure++```+camel-research-assistant/+โโโ app.py # Main Streamlit application+โโโ requirements.txt # Python dependencies+โโโ .env # Environment variables (not included in version control)+โโโ README.md # Project documentation+```++## ๐งช Example Usage++1. Launch the application.+2. Input a topic, e.g., "Latest breakthroughs in quantum computing".+3. Click "Generate Report".+4. The application will:+ - Identify top researchers in the field.+ - Retrieve relevant academic papers and news articles.+ - Generate a comprehensive report.+ - Save the report locally.+ - Optionally, generate illustrative images and prepare content for LinkedIn dissemination.\ No newline at end of filediff --git a/examples/usecases/multi_agent_research_assistant/app.py b/examples/usecases/multi_agent_research_assistant/app.pynew file mode 100644index 000000000..518f8baa7--- /dev/null+++ b/examples/usecases/multi_agent_research_assistant/app.py@@ -0,0 +1,172 @@+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========+# Licensed under the Apache License, Version 2.0 (the "License");+# you may not use this file except in compliance with the License.+# You may obtain a copy of the License at+#+# http://www.apache.org/licenses/LICENSE-2.0+#+# Unless required by applicable law or agreed to in writing, software+# distributed under the License is distributed on an "AS IS" BASIS,+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.+# See the License for the specific language governing permissions and+# limitations under the License.+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========++import streamlit as st+from dotenv import load_dotenv+from camel.societies import RolePlaying+from camel.models import ModelFactory+from camel.types import ModelPlatformType, ModelType+from camel.toolkits import (+ GoogleScholarToolkit, SemanticScholarToolkit, ArxivToolkit,+ AskNewsToolkit, ThinkingToolkit,+ FileWriteToolkit, LinkedInToolkit+)+from camel.logger import set_log_level+import logging++logging.basicConfig(level=logging.DEBUG)+load_dotenv()+set_log_level(level="DEBUG")++# Model Setup+model = ModelFactory.create(+ model_platform=ModelPlatformType.OPENAI,+ model_type=ModelType.GPT_4O,+ model_config_dict={"temperature": 0.0}+)++class DynamicResearchAgent:+ """+ Research agent that can dynamically add GoogleScholar tools based on discovered authors+ """+ def __init__(self, base_tools):+ self.base_tools = base_tools+ self.google_scholar_tools = {}++ def add_google_scholar_for_author(self, author_identifier: str, author_name: str = None):+ """Add GoogleScholar tools for a specific author"""+ if author_identifier not in self.google_scholar_tools:+ try:+ toolkit = GoogleScholarToolkit(author_identifier=author_identifier)+ self.google_scholar_tools[author_identifier] = toolkit.get_tools()+ return True+ except Exception as e:+ st.warning(f"Could not create GoogleScholar toolkit for {author_name or author_identifier}: {e}")+ return False+ return True++ def get_all_tools(self):+ """Get all available tools including dynamically added ones"""+ all_tools = list(self.base_tools)+ for tools in self.google_scholar_tools.values():+ all_tools.extend(tools)+ return all_tools++# Streamlit UI+st.title("CAMEL Dynamic Research Assistant")+topic = st.text_input("Enter a research topic:", value="latest breakthroughs in quantum computing")++if st.button("Generate Report") and topic:+ st.info("๐ค Starting research agent...")++ # Base tools that don't require author identifiers+ base_tools = [+ *SemanticScholarToolkit().get_tools(),+ *ArxivToolkit().get_tools(),+ *AskNewsToolkit().get_tools(),+ *ThinkingToolkit().get_tools(),+ *FileWriteToolkit().get_tools(),+ *LinkedInToolkit().get_tools(),+ ]++ # Create dynamic research agent+ research_agent = DynamicResearchAgent(base_tools)++ # Enhanced task prompt that explains available capabilities+ task_prompt = f"""+ Create a comprehensive research report on: {topic}++ Your complete task includes:+ 1. Search for recent and relevant papers using SemanticScholar and ArXiv+ 2. Identify key researchers and their contributions in this field+ 3. If you find important authors, mention their IDs for potential GoogleScholar integration+ 4. Analyze the latest trends, breakthroughs, and developments+ 5. Get recent news using AskNews if relevant+ 6. Synthesize ALL findings into a well-structured comprehensive report+ 7. Save the final report as a local file using FileWrite tools+ 8. When the report is complete and saved, respond with "CAMEL_TASK_DONE"++ Available tools:+ - SemanticScholar: Search academic papers, get author information+ - ArXiv: Search preprints and recent papers+ - AskNews: Get recent news and developments+ - Thinking: Plan and reflect on your research strategy+ - FileWrite: Save your findings and reports+ - LinkedIn: Research author profiles if needed++ IMPORTANT: Don't just list papers - create a comprehensive analysis report that synthesizes+ the information, identifies trends, and provides insights. Save this report to a file.+ """++ # Initialize RolePlaying session+ role_play = RolePlaying(+ assistant_role_name="Senior Research Analyst",+ user_role_name="Research Director",+ assistant_agent_kwargs=dict(+ model=model,+ tools=research_agent.get_all_tools()+ ),+ user_agent_kwargs=dict(model=model),+ task_prompt=task_prompt,+ with_task_specify=False+ )++ # Start conversation+ next_msg = role_play.init_chat()++ # Conversation loop with dynamic tool addition+ conversation_container = st.container()+ step_count = 0++ with conversation_container:+ while True:+ step_count += 1++ with st.expander(f"Step {step_count}: Agent Interaction", expanded=True):+ assistant_resp, user_resp = role_play.step(next_msg)++ if assistant_resp.terminated or user_resp.terminated:+ st.info("๐ Conversation terminated by agent")+ break++ # Check if agent mentions needing GoogleScholar for specific authors+ # This is a simple pattern - you could make this more sophisticated+ content = assistant_resp.msg.content.lower()+ if "google scholar" in content and "author" in content:+ st.info("๐ Agent requested GoogleScholar tools - this could be implemented with author discovery")++ # Display conversation+ st.markdown("**๐ค Research Analyst:**")+ st.write(assistant_resp.msg.content)++ st.markdown("**๐ค Research Director:**")+ st.write(user_resp.msg.content)++ # Check for completion in both agent responses+ if ("CAMEL_TASK_DONE" in user_resp.msg.content or+ "CAMEL_TASK_DONE" in assistant_resp.msg.content or+ "report is complete" in assistant_resp.msg.content.lower() or+ "task completed" in assistant_resp.msg.content.lower()):+ st.success("โ Task completed successfully!")+ break++ next_msg = assistant_resp.msg++ # Safety break to prevent infinite loops+ if step_count > 20:+ st.warning("โ ๏ธ Maximum steps reached. Stopping conversation.")+ break++ st.success("๐ Report generation completed!")+ st.info("๐ The research report has been saved locally by the agent.")\ No newline at end of filediff --git a/examples/usecases/multi_agent_research_assistant/requirements.txt b/examples/usecases/multi_agent_research_assistant/requirements.txtnew file mode 100644index 000000000..c913a7079--- /dev/null+++ b/examples/usecases/multi_agent_research_assistant/requirements.txt@@ -0,0 +1,6 @@+chunkr-ai==0.1.3+python-dotenv==1.1.0+streamlit==1.5.1+camel-ai[all]==0.2.62+docx2markdown==0.1.0+asknews==0.11.10\ No newline at end of file