Notebook

A scratchpad the agent can read and write across invocations. The most effective use is giving the agent a notebook at the start of a task and instructing it to plan its work there. It can break the task into steps, check things off as it goes, and always have a clear picture of what’s left. Notebook state is part of the agent’s state, so it persists automatically with Session Management.

Supported in: Node.js, modern browsers (TypeScript); all platforms (Python).

Example - Task Management:

TypeScript
Python

import { Agent } from '@strands-agents/sdk'
import { notebook } from '@strands-agents/sdk/vended-tools/notebook'

const agent = new Agent({
  tools: [notebook],
  systemPrompt:
    'Before starting any multi-step task, create a notebook with a checklist of steps. ' +
    'Check off each step as you complete it.',
})

// The agent uses the notebook to plan and track its work
await agent.invoke('Write a project plan for building a personal budget tracker app')

from strands import Agent
from strands.vended_tools import notebook

agent = Agent(tools=[notebook])
agent('Create a notebook called "tasks" with "# Daily Tasks" and add "- [ ] Review code" to it')

Example - State Persistence:

TypeScript
Python

import { Agent, SessionManager, FileStorage } from '@strands-agents/sdk'
import { notebook } from '@strands-agents/sdk/vended-tools/notebook'

const session = new SessionManager({
  sessionId: 'my-session',
  storage: { snapshot: new FileStorage('./sessions') },
})

const agent = new Agent({ tools: [notebook], sessionManager: session })

// Notebooks are automatically persisted as part of the session
await agent.invoke('Create a notebook called "ideas" with "# Project Ideas"')
await agent.invoke('Add "- Build a web scraper" to the ideas notebook')

// ...

// Later, a new agent with the same session restores notebooks automatically
const restoredAgent = new Agent({ tools: [notebook], sessionManager: session })
await restoredAgent.invoke('Read the ideas notebook')

from strands import Agent
from strands.vended_tools import notebook

agent = Agent(tools=[notebook])
agent('Create a notebook called "tasks" with "# Daily Tasks" and add "- [ ] Review code" to it')

# Read the notebook contents directly off agent state.
notebooks = agent.state.get("notebooks") or {}
print(notebooks.get("tasks"))

Example - Custom configuration:

TypeScript
Python

import { Agent } from '@strands-agents/sdk'
import { makeNotebook } from '@strands-agents/sdk/vended-tools/notebook'

const notes = makeNotebook({
  name: 'notes',
  maxNotebookSizeBytes: 64 * 1024, // 64 KiB
})
const agent = new Agent({ tools: [notes] })

from strands import Agent
from strands.vended_tools import make_notebook

notes = make_notebook(
    name="notes",
    max_notebook_size_bytes=64 * 1024,  # 64 KiB
)
agent = Agent(tools=[notes])

📖 Full API Reference