SuperAPI AI Directory

dexter

Autonomous financial research agent designed to perform deep analysis by thinking, planning, and learning through iteration. Specialized for the financial domain.

Open website

About this tool

Dexter is an autonomous financial research agent that can think, plan, and iteratively learn like Claude Code, built specifically for the financial research domain. It completes analysis through task planning, self-reflection, and real-time market data, transforming complex financial questions into clear step-by-step research plans, and autonomously executes, checks its own work, and refines results until it reaches reliable, data-backed answers.

Main capabilities:

  • Intelligent task planning: automatically breaks down complex queries into structured research steps.
  • Autonomous execution: selects and executes appropriate tools to collect financial data.
  • Self-verification: checks its own work and iterates until the task is complete.
  • Real-time financial data: has access to income statements, balance sheets, and cash flow statements.
  • Safety features: built-in loop detection and step limits to prevent runaway execution.

How to use:

  1. Prerequisites: requires the Bun runtime (v1.0 or higher), an OpenAI API key, a Financial Datasets API key, and an optional Exa API key (for web search).
  2. Install Bun (if not installed): install using the curl command, then restart the terminal and verify after installation.
  3. Clone the repository and install dependencies: install dependencies using Bun.
  4. Set environment variables.
  5. Run Dexter: interactive mode or development watch mode.

Evaluation: Dexter includes an evaluation suite, uses LangSmith for tracing, and employs an LLM as a judge to score correctness. You can run all questions or a random sample, and the evaluation runner displays real-time progress, the current question, and accuracy statistics.

Logging: Dexter logs all tool calls to scratch files (JSONL files in the .dexter/scratchpad/ directory) for easy debugging and historical tracking. Each file contains the original query, tool call results, and the agent's reasoning steps.

WhatsApp integration: You can chat with Dexter in WhatsApp by linking your phone to the gateway. Messages sent to yourself are processed by Dexter, and replies are sent back to the same chat. For detailed setup, see the WhatsApp Gateway README.

Contributing: You are welcome to fork the repository, create a feature branch, commit changes, push, and create a Pull Request. Please keep PRs small and focused.

Limitations and disclaimer: This project is for educational, entertainment, and informational purposes only and is not suitable for real trading or investing. It does not constitute financial, investment, tax, or legal advice. Accuracy, completeness, or suitability is not guaranteed. Outputs may be incorrect, incomplete, or outdated. The creators and contributors are not liable for any financial loss or damage. Consult a licensed financial advisor before making investment decisions. Past performance does not indicate future results. By using this software, you agree to use it only for learning and informational purposes and to assume all related risks.

License: MIT License.