Social Research Agent

Social Research is a Python application that uses an AI agent to research a topic across many social and news platforms, then delivers a cited markdown report and a live analytics dashboard. What it does You enter a topic (and optionally a time window and platforms). The system: 1. Runs a Gemini-driven agent loop that chooses platforms, refines queries, and calls tools until it decides it’s done. 2. Fetches posts from adapters (Reddit, Hacker News, Mastodon, Bluesky, Stack Exchange, Lemmy, Google News, Google Trends, YouTube, and optional scrape-based X/TikTok). 3. Updates analytics incrementally — volume, sentiment, communities, top authors, hot posts, subtopic clusters. 4. Streams progress to the browser via Server-Sent Events (timeline of agent steps). 5. Persists sessions, posts, events, and the final report in SQLite (SQLAlchemy 2.0 async + Alembic). The design is inspired by multi-source research agents (e.g. “local-deep-research”) but focused on social discourse and trends, not academic papers.