Core

Agent

The AI agent that powers intelligent assistance in Thesis Notebook.

Overview

The Thesis Agent is an AI-powered assistant built on OpenAI's agent framework. It can understand your notebook context, answer questions, plan complex tasks, and execute multi-step operations autonomously.

Key capabilities:

  • Read and understand notebook content and structure
  • Create, edit, and execute notebook cells
  • Search the web for documentation and solutions
  • Create structured plans for complex tasks
  • Execute shell commands for file operations

How It Works

When you interact with the agent, it follows this process:

  1. Context gathering - The agent receives your message along with relevant notebook context
  2. Planning - For complex tasks, it creates a plan of action
  3. Execution - It uses tools to accomplish the task, iterating as needed
  4. Response - Results are streamed back in real-time

The agent can run for up to 5000 turns, allowing it to handle complex multi-step workflows.

Agent Modes

The agent operates in three modes with different capabilities and tool access:

ModeToolsUse Case
Ask4 toolsRead-only exploration and questions
Plan5 toolsResearch and task planning
Agent13 toolsFull autonomous execution

Learn more about agent modes →

Available Tools

The agent has access to various tools depending on the mode:

  • File operations - Read files, list directories
  • Notebook operations - Read, create, edit, run cells
  • Web search - Search for documentation and solutions
  • Planning - Create structured task plans
  • Shell execution - Run system commands

See the complete tool reference →