An autonomous agent that thinks, plans, and learns to perform deep financial analysis using task planning, self-validation, and real-time market data.

🎯 Core Philosophy: Think, Plan, Learn, Answer

Dexter is designed to act like a specialized financial analyst. It doesn't just retrieve data; it understands a query, devises a multi-step research plan, executes it using various tools, checks its work for completeness, and synthesizes a final, well-reasoned answer.

✨ Key Features & Capabilities

🧠 Intelligent Task Planning: Automatically decomposes complex financial questions into structured, actionable research steps.

🤖 Autonomous Execution: Selects and runs the appropriate tools to gather necessary financial data (income statements, balance sheets, cash flows, etc.).

🔍 Self-Validation & Reflection: Has a built-in validation agent to check its own work, ensuring data sufficiency and task completion before finalizing answers.

📈 Real-Time Financial Data: Integrates with live financial datasets for up-to-date analysis.

🛡️ Safety & Control: Includes features like loop detection and step limits to prevent infinite or runaway execution cycles.

🚀 Quick Start Guide

Prerequisites

Bun Runtime (v1.0+): The primary JavaScript/TypeScript runtime.

API Keys: You'll need at minimum an OpenAI API key and a Financial Datasets API key. Optional keys for xAI (Grok), Anthropic (Claude), Google (Gemini), and Tavily (web search) provide more model choices.

Installation & Setup

1. Install Bun (if not installed):

BASH

# macOS/Linux

curl -fsSL https://bun.com/install | bash

2. Clone and Setup Dexter:

BASH

git clone https://github.com/virattt/dexter.git

cd dexter bun install cp env.example .env # Edit `.env` and add your API keys

3. Run Dexter:

BASH

bun start # For interactive mode

# OR bun dev # For development/watch mode

💡 Example Queries

Ask Dexter questions like:

"What was Apple's revenue growth over the last 4 quarters?"

"Compare Microsoft and Google's operating margins for 2023."

"Analyze Tesla's cash flow trends over the past year."

"What is Amazon's debt-to-equity ratio based on recent financials?"

Dexter will handle the entire research workflow from planning to final answer.

🏗️ Technical Architecture

Dexter uses a multi-agent system:

Planning Agent: Analyzes the query and creates a structured task list.

Action Agent: Selects tools and executes the research steps.

Validation Agent: Verifies task completion and data quality.

Answer Agent: Synthesizes all findings into a comprehensive response.

🛠️ Tech Stack

Runtime: Bun

Language: TypeScript

UI: React + Ink (for the terminal interface)

LLM Framework: LangChain.js with multi-provider support (OpenAI, Anthropic, Google, xAI, Ollama)

Validation: Zod

Model Flexibility: Type /model in the CLI to switch between GPT-4.1, Claude 3.5 Sonnet, Gemini 3, Grok 4, or local Ollama models.

📊 Repository Info

Primary Language: TypeScript (99.2%)

License: MIT

Status: Actively developed (latest commits in Jan 2026).

About

Dexter is an autonomous AI agent built for deep financial research. Think of it as a specialized "Claude Code" for finance. It takes complex financial questions, breaks them down into structured plans, autonomously executes research tasks using real-time market data, validates its own findings, and delivers comprehensive, data-backed answers.

58 files

9 folders

328.27 KB total size

0 open issues

0 open pull requests

0 watchers

0 forks

0 stars

97 views

Updated Jan 20, 2026

Recent Commits View all

Update README.md

WebDev committed Jan 20, 2026

Initial commit - Upload project 'dexter'

WebDev committed Jan 20, 2026

Languages

TypeScript 99.2%

JavaScript 0.8%