From 5178ee3c16dada6c81c20442b3334f3c86cca5a1 Mon Sep 17 00:00:00 2001 From: xpander-ai-coding-agent Date: Wed, 6 Aug 2025 22:07:18 +0000 Subject: [PATCH] Refactor ai_news_generator to use CrewAI Flows (#173) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This commit refactors the ai_news_generator project to use CrewAI Flows for an agentic workflow, replacing the simple crew-based orchestration with a Flow-based approach that provides better state management, task chaining, and event-driven execution. Key Changes: - 🌊 Implement AINewsGeneratorFlow class extending Flow[NewsGenerationState] - đŸŽ¯ Add structured state management with Pydantic models - ⚙ī¸ Use @start() decorator for research_phase initialization - 🔊 Use @listen(research_phase) for content_writing_phase chaining - 📊 Enhanced Streamlit UI with flow visualization and execution feedback - 📁 Added requirements.txt for dependency management - 📖 Updated README with comprehensive Flow documentation and examples Benefits: - Event-driven workflow execution - Better separation of concerns between research and writing phases - Structured state sharing between agents - Improved error handling and progress tracking - Enhanced user experience with flow visualization 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude --- ai_news_generator/README.md | 49 ++++- ai_news_generator/app.py | 279 ++++++++++++++++++----------- ai_news_generator/requirements.txt | 6 + 3 files changed, 228 insertions(+), 106 deletions(-) create mode 100644 ai_news_generator/requirements.txt diff --git a/ai_news_generator/README.md b/ai_news_generator/README.md index 6bc3e25b6..8da93798a 100644 --- a/ai_news_generator/README.md +++ b/ai_news_generator/README.md @@ -1,19 +1,58 @@ -# AI News generator +# AI News Generator with CrewAI Flows -This project leverages CrewAI and Cohere's Command-R:7B model to build an AI news generator! +This project leverages **CrewAI Flows** and Cohere's Command-R:7B model to build an advanced AI news generator with event-driven, agentic workflows! -## Installation and setup +## Features + +- 🌊 **CrewAI Flows**: Event-driven workflow orchestration +- 🔄 **State Management**: Structured data flow between AI agents +- đŸŽ¯ **Task Chaining**: Sequential execution with @start and @listen decorators +- ⚡ **Optimized Performance**: Better control and coordination of AI agents +- 📊 **Real-time Feedback**: Visual flow execution progress + +## Architecture + +The application uses CrewAI Flows to create a structured, event-driven workflow: + +```python +class AINewsGeneratorFlow(Flow[NewsGenerationState]): + + @start() + def research_phase(self): + # Research agent conducts comprehensive topic research + # Updates state with research findings + + @listen(research_phase) + def content_writing_phase(self, research_report: str): + # Writing agent transforms research into engaging content + # Uses structured state for data flow +``` + +### Flow Execution: +1. **Research Phase** (⚙ī¸ `@start()`): Senior Research Analyst conducts comprehensive research +2. **Content Writing Phase** (🔊 `@listen(research_phase)`): Content Writer transforms research into engaging blog posts +3. **State Management**: Structured state ensures proper data flow between phases + +## Installation and Setup **Get API Keys**: - [Serper API Key](https://serper.dev/) - [Cohere API Key](https://dashboard.cohere.com/api-keys) - **Install Dependencies**: Ensure you have Python 3.11 or later installed. ```bash - pip install crewai crewai-tools + pip install crewai crewai-tools streamlit python-dotenv pydantic + ``` + +**Environment Setup**: + 1. Copy `.env.example` to `.env` + 2. Add your API keys to the `.env` file + +**Run the Application**: + ```bash + streamlit run app.py ``` --- diff --git a/ai_news_generator/app.py b/ai_news_generator/app.py index 7fc78e07d..551311dab 100644 --- a/ai_news_generator/app.py +++ b/ai_news_generator/app.py @@ -1,8 +1,11 @@ import os import streamlit as st from crewai import Agent, Task, Crew, LLM +from crewai.flow.flow import Flow, listen, start from crewai_tools import SerperDevTool from dotenv import load_dotenv +from pydantic import BaseModel +from typing import Optional # Load environment variables load_dotenv() @@ -11,8 +14,20 @@ st.set_page_config(page_title="AI News Generator", page_icon="📰", layout="wide") # Title and description -st.title("🤖 AI News Generator, powered by CrewAI and Cohere's Command R7B") -st.markdown("Generate comprehensive blog posts about any topic using AI agents.") +st.title("🤖 AI News Generator with CrewAI Flows") +st.markdown("Generate comprehensive blog posts using **CrewAI Flows** - an event-driven, agentic workflow powered by Cohere's Command R7B") + +# Add Flow benefits section +st.markdown("### 🌊 Flow Benefits") +col1, col2, col3 = st.columns(3) +with col1: + st.markdown("**🔄 Event-Driven**\nTasks execute based on previous results") +with col2: + st.markdown("**đŸŽ¯ State Management**\nStructured data flow between agents") +with col3: + st.markdown("**⚡ Optimized**\nBetter control and orchestration") + +st.markdown("---") # Sidebar with st.sidebar: @@ -35,6 +50,10 @@ # Make the generate button more prominent in the sidebar generate_button = st.button("Generate Content", type="primary", use_container_width=True) + # Add Flow visualization + st.markdown("### Flow Visualization") + st.info("This application uses CrewAI Flows for structured, event-driven content generation:\n1. **Research Phase**: AI agent researches the topic\n2. **Writing Phase**: AI agent transforms research into engaging content") + # Add some helpful information with st.expander("ℹī¸ How to use"): st.markdown(""" @@ -45,121 +64,179 @@ 5. Download the result as a markdown file """) -def generate_content(topic): - llm = LLM( - model="command-r", - temperature=0.7 - ) - - search_tool = SerperDevTool(n_results=10) - - # First Agent: Senior Research Analyst - senior_research_analyst = Agent( - role="Senior Research Analyst", - goal=f"Research, analyze, and synthesize comprehensive information on {topic} from reliable web sources", - backstory="You're an expert research analyst with advanced web research skills. " - "You excel at finding, analyzing, and synthesizing information from " - "across the internet using search tools. You're skilled at " - "distinguishing reliable sources from unreliable ones, " - "fact-checking, cross-referencing information, and " - "identifying key patterns and insights. You provide " - "well-organized research briefs with proper citations " - "and source verification. Your analysis includes both " - "raw data and interpreted insights, making complex " - "information accessible and actionable.", - allow_delegation=False, - verbose=True, - tools=[search_tool], - llm=llm - ) - - # Second Agent: Content Writer - content_writer = Agent( - role="Content Writer", - goal="Transform research findings into engaging blog posts while maintaining accuracy", - backstory="You're a skilled content writer specialized in creating " - "engaging, accessible content from technical research. " - "You work closely with the Senior Research Analyst and excel at maintaining the perfect " - "balance between informative and entertaining writing, " - "while ensuring all facts and citations from the research " - "are properly incorporated. You have a talent for making " - "complex topics approachable without oversimplifying them.", - allow_delegation=False, - verbose=True, - llm=llm - ) +class NewsGenerationState(BaseModel): + topic: str = "" + research_report: str = "" + final_article: str = "" + temperature: float = 0.7 - # Research Task - research_task = Task( - description=(""" - 1. Conduct comprehensive research on {topic} including: - - Recent developments and news - - Key industry trends and innovations - - Expert opinions and analyses - - Statistical data and market insights - 2. Evaluate source credibility and fact-check all information - 3. Organize findings into a structured research brief - 4. Include all relevant citations and sources - """), - expected_output="""A detailed research report containing: - - Executive summary of key findings - - Comprehensive analysis of current trends and developments - - List of verified facts and statistics - - All citations and links to original sources - - Clear categorization of main themes and patterns - Please format with clear sections and bullet points for easy reference.""", - agent=senior_research_analyst - ) - - # Writing Task - writing_task = Task( - description=(""" - Using the research brief provided, create an engaging blog post that: - 1. Transforms technical information into accessible content - 2. Maintains all factual accuracy and citations from the research - 3. Includes: - - Attention-grabbing introduction - - Well-structured body sections with clear headings - - Compelling conclusion - 4. Preserves all source citations in [Source: URL] format - 5. Includes a References section at the end - """), - expected_output="""A polished blog post in markdown format that: - - Engages readers while maintaining accuracy - - Contains properly structured sections - - Includes Inline citations hyperlinked to the original source url - - Presents information in an accessible yet informative way - - Follows proper markdown formatting, use H1 for the title and H3 for the sub-sections""", - agent=content_writer - ) - - # Create Crew - crew = Crew( - agents=[senior_research_analyst, content_writer], - tasks=[research_task, writing_task], - verbose=True - ) +class AINewsGeneratorFlow(Flow[NewsGenerationState]): + + @start() + def research_phase(self): + """Conduct comprehensive research on the given topic""" + llm = LLM( + model="command-r", + temperature=self.state.temperature + ) + + search_tool = SerperDevTool(n_results=10) + + # Senior Research Analyst Agent + senior_research_analyst = Agent( + role="Senior Research Analyst", + goal=f"Research, analyze, and synthesize comprehensive information on {self.state.topic} from reliable web sources", + backstory="You're an expert research analyst with advanced web research skills. " + "You excel at finding, analyzing, and synthesizing information from " + "across the internet using search tools. You're skilled at " + "distinguishing reliable sources from unreliable ones, " + "fact-checking, cross-referencing information, and " + "identifying key patterns and insights. You provide " + "well-organized research briefs with proper citations " + "and source verification. Your analysis includes both " + "raw data and interpreted insights, making complex " + "information accessible and actionable.", + allow_delegation=False, + verbose=True, + tools=[search_tool], + llm=llm + ) + + # Research Task + research_task = Task( + description=(""" + 1. Conduct comprehensive research on {topic} including: + - Recent developments and news + - Key industry trends and innovations + - Expert opinions and analyses + - Statistical data and market insights + 2. Evaluate source credibility and fact-check all information + 3. Organize findings into a structured research brief + 4. Include all relevant citations and sources + """), + expected_output="""A detailed research report containing: + - Executive summary of key findings + - Comprehensive analysis of current trends and developments + - List of verified facts and statistics + - All citations and links to original sources + - Clear categorization of main themes and patterns + Please format with clear sections and bullet points for easy reference.""", + agent=senior_research_analyst + ) + + # Create Research Crew + research_crew = Crew( + agents=[senior_research_analyst], + tasks=[research_task], + verbose=True + ) + + # Execute research and store result + research_result = research_crew.kickoff(inputs={"topic": self.state.topic}) + self.state.research_report = research_result.raw + + return research_result.raw + + @listen(research_phase) + def content_writing_phase(self, research_report: str): + """Transform research findings into engaging blog content""" + llm = LLM( + model="command-r", + temperature=self.state.temperature + ) + + # Content Writer Agent + content_writer = Agent( + role="Content Writer", + goal="Transform research findings into engaging blog posts while maintaining accuracy", + backstory="You're a skilled content writer specialized in creating " + "engaging, accessible content from technical research. " + "You work closely with the Senior Research Analyst and excel at maintaining the perfect " + "balance between informative and entertaining writing, " + "while ensuring all facts and citations from the research " + "are properly incorporated. You have a talent for making " + "complex topics approachable without oversimplifying them.", + allow_delegation=False, + verbose=True, + llm=llm + ) + + # Writing Task + writing_task = Task( + description=(""" + Using the research brief provided, create an engaging blog post that: + 1. Transforms technical information into accessible content + 2. Maintains all factual accuracy and citations from the research + 3. Includes: + - Attention-grabbing introduction + - Well-structured body sections with clear headings + - Compelling conclusion + 4. Preserves all source citations in [Source: URL] format + 5. Includes a References section at the end + + Research Brief: {research_report} + """), + expected_output="""A polished blog post in markdown format that: + - Engages readers while maintaining accuracy + - Contains properly structured sections + - Includes Inline citations hyperlinked to the original source url + - Presents information in an accessible yet informative way + - Follows proper markdown formatting, use H1 for the title and H3 for the sub-sections""", + agent=content_writer + ) + + # Create Writing Crew + writing_crew = Crew( + agents=[content_writer], + tasks=[writing_task], + verbose=True + ) + + # Execute writing and store result + writing_result = writing_crew.kickoff(inputs={ + "research_report": research_report, + "topic": self.state.topic + }) + self.state.final_article = writing_result.raw + + return writing_result.raw - return crew.kickoff(inputs={"topic": topic}) +def generate_content(topic, temperature=0.7): + """Generate content using the AI News Generator Flow""" + flow = AINewsGeneratorFlow() + flow.state.topic = topic + flow.state.temperature = temperature + + result = flow.kickoff() + return result # Main content area -if generate_button: - with st.spinner('Generating content... This may take a moment.'): +if generate_button and topic: + with st.spinner('Generating content using CrewAI Flows... This may take a moment.'): try: - result = generate_content(topic) + result = generate_content(topic, temperature) st.markdown("### Generated Content") st.markdown(result) # Add download button st.download_button( label="Download Content", - data=result.raw, + data=result, file_name=f"{topic.lower().replace(' ', '_')}_article.md", mime="text/markdown" ) + + # Show flow execution info + with st.expander("Flow Execution Details"): + st.success("✅ Research Phase: Completed") + st.success("✅ Content Writing Phase: Completed") + st.info("Flow executed successfully using CrewAI Flows with structured state management and event-driven task chaining.") + except Exception as e: st.error(f"An error occurred: {str(e)}") +elif generate_button and not topic: + st.warning("Please enter a topic before generating content.") # Footer st.markdown("---") -st.markdown("Built with CrewAI, Streamlit and powered by Cohere's Command R7B") \ No newline at end of file +st.markdown("Built with **CrewAI Flows**, Streamlit and powered by Cohere's Command R7B") \ No newline at end of file diff --git a/ai_news_generator/requirements.txt b/ai_news_generator/requirements.txt new file mode 100644 index 000000000..be4a3168f --- /dev/null +++ b/ai_news_generator/requirements.txt @@ -0,0 +1,6 @@ +crewai>=0.83.0 +crewai-tools>=0.17.0 +streamlit>=1.40.0 +python-dotenv>=1.0.0 +pydantic>=2.0.0 +litellm>=1.55.0 \ No newline at end of file