kingabzpro/Multi-Agent-Research-Assistant

Multi-Agent Research Assistant: a developer tool that coordinates multiple LLM agents to turn a user research question into a sourced, polished Markdown report rendered in the browser and exportable as a PDF. Built with the OpenAI Agents SDK, Olostep (Answer/Search/Scrape), and Reflex. Who uses it: researchers, engineers, or knowledge workers who need source-backed summaries and reproducible retrieval traces. Key features: - Three-agent workflow (Manager orchestrates, Judge evaluates evidence, Analyst writes the report). - Staged retrieval policy: Olostep Answer → Judge (threshold 0.85) → Search+Scrape → targeted searches and scraping of top 3 pages if needed. - Live progress logs in the Reflex UI, styled Markdown rendering, and ReportLab PDF export. - Code layout: Reflex UI (app/app.py), state/handlers (app/state.py), agents/workflow (app/research_assistant.py), formatting (app/report_formatting.py), PDF export (app/pdf_export.py). Setup via .env and pip requirements.

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。