Epic News Crew

Natural-language AI assistant for financial analysis, news, research, recipes, travel, OSINT, and more. Built on crewAI: ask in plain English or French and the right team of AI agents (a "crew") handles your request.

TL;DR

pip install uv
uv sync && uv pip install -e .
cp .env.example .env       # then add your API keys
crewai flow kickoff        # type a request when prompted

Reports land in output/<crew>/ and (if Gmail is connected via Composio) in your inbox.

Documentation

📖 Browse the docs site: fjacquet.github.io/epic_news — auto-published from docs/ on every push to master.

The source markdown lives in docs/, organized in the Diátaxis style.

If you want to…	Go to
Discover what you can ask the system	User Guide
Build your first crew	Tutorial: Your First Crew
Set up the dev environment	Development Setup
Understand the architecture	Architecture
Browse Architecture Decision Records	docs/adr/
See what changed	CHANGELOG.md

Project Highlights

• ~25 specialized crews — finance, news, OSINT, cooking, legal, sales, travel, deep research, PESTEL…
• Real-time data over RAG — agents call live tools (Perplexity, Tavily, YahooFinance, …) per execution
• DOCX reports — Pydantic models → one assembler per crew → pandoc → a .docx file under output/, attached to the email and offered for download in the app
• Composable observability — loguru trace per run, structured PostResult for email outcomes

Contributing

PRs welcome. Please run make validate (lint + tests) before opening one. See the Development Setup for the full workflow.

Support

• crewAI documentation
• Issue tracker