The latest articles on DEV Community by Jaeil Woo (@jiwoomap). https://dev.to/jiwoomap https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3705060%2Ff85c9e90-0fa5-43d0-b655-25e88d4137ec.png https://dev.to/jiwoomap en Jaeil Woo Sun, 11 Jan 2026 09:52:16 +0000 https://dev.to/jiwoomap/building-a-remembering-ai-trading-agent-with-python-langgraph-and-obsidian-30hn https://dev.to/jiwoomap/building-a-remembering-ai-trading-agent-with-python-langgraph-and-obsidian-30hn <p>Hello DEV community!</p> <p>I'm excited to share an open-source project I've been working on: <strong>TradingAgents-Dashboard</strong>.<br> It's a Dockerized AI trading assistant that not only analyzes the market but <strong>remembers your insights forever</strong> using a local knowledge base (RAG).</p> <h2> The Problem: "Stateless" AI </h2> <p>Most AI trading bots today are "stateless". They run an analysis, give you a result, and then <strong>forget everything</strong> the moment you close the terminal.</p> <ul> <li> "Wait, didn't we decide last week that inflation correlates with this stock?"</li> <li> "Where is that news link I saw yesterday?"</li> </ul> <p>As a developer and trader, I wanted an agent that <strong>grows smarter</strong> over time, just like a human analyst.</p> <h2> The Solution: AI + Obsidian (RAG) </h2> <p>I built a dashboard wrapping the <a href="https://github.com/TauricResearch/TradingAgents" rel="noopener noreferrer">TradingAgents</a> framework, adding a persistent memory layer using <strong>Obsidian</strong>.</p> <p><strong>Github Repo:</strong> <a href="https://github.com/jiwoomap/TradingAgents-Dashboard" rel="noopener noreferrer">jiwoomap/TradingAgents-Dashboard</a></p> <h3> How it works: </h3> <ol> <li> <strong>Analyze:</strong> Agents (Bull, Bear, Risk Manager) debate market conditions using LangGraph.</li> <li> <strong>Persist:</strong> All insights and debates are auto-saved to your local <strong><a href="https://obsidian.md" rel="noopener noreferrer">Obsidian Vault</a></strong> as Markdown files.</li> <li> <strong>Recall (RAG):</strong> Before making a new decision, the agents search your vault (via ChromaDB) to retrieve past lessons and context.</li> </ol> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Foe2imql6s1bubc1if40u.gif" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Foe2imql6s1bubc1if40u.gif" alt="Demo Animation" width="600" height="324"></a></p> <h2> Tech Stack </h2> <ul> <li> <strong>Framework:</strong> LangChain / LangGraph (Multi-Agent Orchestration)</li> <li> <strong>UI:</strong> Streamlit (Web Dashboard)</li> <li> <strong>Database:</strong> ChromaDB (Vector Store for RAG)</li> <li> <strong>Memory:</strong> Obsidian (Markdown-based Knowledge Base)</li> <li> <strong>Infrastructure:</strong> Docker &amp; Docker Compose</li> </ul> <h2> Key Features </h2> <ul> <li> <strong>Interactive Debate UI:</strong> Watch the "Bull" and "Bear" agents fight it out in real-time.</li> <li> <strong>Fact Checker:</strong> Prevents hallucinations by validating news URLs (200 OK checks).</li> <li> <strong>Dockerized:</strong> Get started in 1 minute with <code>docker-compose up</code>.</li> <li> <strong>Data Sovereignty:</strong> Your financial data and strategies live on <strong>your disk</strong>, not in a cloud database.</li> </ul> <h2> Try it out! </h2> <p>I'd love to get your feedback. If you're interested in AI Agents or FinTech, give it a spin!</p> <ol> <li> Clone the repo: git clone <a href="https://github.com/jiwoomap/TradingAgents-Dashboard.git" rel="noopener noreferrer">https://github.com/jiwoomap/TradingAgents-Dashboard.git</a> </li> </ol> opensource python machinelearning ai