How jahnavikorukonda/multi_agent_system works
ResearchMind accepts a topic, searches for sources, scrapes a relevant page, drafts a report, and critiques it. The Streamlit app orchestrates and displays the interactive workflow; pipeline.py offers a separate command-line workflow. The sampled code supports the search, scrape, writing, and critique stages, but does not show persistence or other result destinations.
Main entry points: agents.py Agents and chains · app.py Streamlit app · pipeline.py CLI pipeline · tools.py Search and scrape tools
Components 12
Entry points
- Streamlit app
app.py - CLI pipeline
pipeline.py
Research stages
- Agents and chains
agents.py - Search and scrape tools
tools.py - Web search
tools.py - Page scraping
tools.py - Report writer
agents.py - Report critic
agents.py
External services
- Groq LLM
- Tavily Search
- Source websites
Other components
- Researcher
Connections 17
- Researcher → Streamlit app: enters topic
- Researcher → CLI pipeline: enters topic
- Streamlit app → Agents and chains: builds and invokes
- CLI pipeline → Web search: invokes search
- CLI pipeline → Agents and chains: builds reader; invokes chains
- Agents and chains → Search and scrape tools: configures agent tools
- Search and scrape tools → Web search: defines
- Search and scrape tools → Page scraping: defines
- Web search → Tavily Search: queries
- Page scraping → Source websites: fetches page
- Agents and chains → Groq LLM: sends prompts
- Agents and chains → Report writer: defines chain
- Agents and chains → Report critic: defines chain
- Streamlit app → Report writer: invokes chain
- Streamlit app → Report critic: invokes chain
- Streamlit app → Researcher: displays results
- CLI pipeline → Researcher: returns state