examples/demo-agent-block.deepnote
Rama main · réplica de lectura del repositorio original.
metadata:
createdAt: 2026-03-03T00:00:00.000Z
snapshotHash: sha256:ed2500292c2fca375116484f886beb8da89082d9e58656cb8f71eefc3140d921
modifiedAt: '2026-03-03T13:00:22.842Z'
project:
id: demo-llm-agentic
name: Agentic Analysis Demo
notebooks:
- blocks:
- blockGroup: b0000000000000000000000000000001
content: Quarterly Sales Analysis — Agentic Demo
id: a0000000000000000000000000000001
metadata: {}
sortingKey: a0
type: text-cell-h1
contentHash: sha256:229240314154c38946d7b93c9488c599a46b485516f7bf29ca1426dffcac0c88
- blockGroup: b0000000000000000000000000000002
content: |
This notebook demonstrates the **agent block**. The code below generates
a synthetic sales dataset. Then the agent autonomously explores the data,
creates visualizations, identifies trends, and writes a report — all by adding
new blocks to this notebook.
id: a0000000000000000000000000000002
metadata: {}
sortingKey: a1
type: markdown
contentHash: sha256:31dd46c96b52ead04b0ac37819dd372b39b54ea368f1b5d8ff712a8ebcb1adae
- blockGroup: b0000000000000000000000000000003
content: |
import pandas as pd
import numpy as np
np.random.seed(42)
regions = ['North America', 'Europe', 'Asia Pacific', 'Latin America']
products = ['Enterprise', 'Professional', 'Starter']
quarters = ['Q1 2025', 'Q2 2025', 'Q3 2025', 'Q4 2025', 'Q1 2026']
rows = []
base_revenue = {'Enterprise': 50000, 'Professional': 20000, 'Starter': 5000}
growth_rate = {'North America': 1.08, 'Europe': 1.12, 'Asia Pacific': 1.20, 'Latin America': 1.15}
for qi, q in enumerate(quarters):
for region in regions:
for product in products:
base = base_revenue[product] * (growth_rate[region] ** qi)
revenue = base * np.random.uniform(0.85, 1.15)
deals = int(revenue / (base_revenue[product] * 0.01) * np.random.uniform(0.8, 1.2))
churn = np.random.uniform(0.02, 0.12) if product != 'Starter' else np.random.uniform(0.08, 0.22)
rows.append({
'quarter': q, 'region': region, 'product': product,
'revenue': round(revenue, 2), 'deals_closed': deals,
'churn_rate': round(churn, 4)
})
df = pd.DataFrame(rows)
print(f"Dataset: {len(df)} rows × {len(df.columns)} columns")
print(f"Quarters: {df['quarter'].nunique()} | Regions: {df['region'].nunique()} | Products: {df['product'].nunique()}")
print(f"\nTotal revenue: ${df['revenue'].sum():,.0f}")
print(f"\nSample rows:")
print(df.head(8).to_string(index=False))
id: a0000000000000000000000000000003
metadata: {}
sortingKey: a2
type: code
contentHash: sha256:1c4986dee40c59936363029ae29900156485aa1961ed620c5271d39d86208525
- blockGroup: b0000000000000000000000000000004
content: |
You are a senior data analyst. The code block above created a pandas DataFrame
called `df` with quarterly sales data across regions and products.
Perform a thorough exploratory analysis. You MUST do ALL of the following, each
as a separate code or markdown block:
1. Add a markdown block introducing your analysis plan
2. Create a grouped bar chart of revenue by quarter, colored by region (use matplotlib, save as PNG)
3. Create a line chart showing churn rate trends over quarters for each product tier
4. Compute a pivot table of average revenue by region × product and display it
5. Identify the single best-performing and worst-performing region+product combo and explain why
6. Use the filesystem MCP tool to save a final executive report as `examples/report.md`
7. Add a final markdown block with an executive summary containing 3 key takeaways and a recommendation
Use clean, publication-quality matplotlib charts with proper labels and titles.
Use `plt.tight_layout()` and `plt.savefig()` for each chart.
id: a0000000000000000000000000000004
metadata:
deepnote_agent_model: auto
deepnote_mcp_servers:
- name: filesystem
command: npx
args:
- '-y'
- '@modelcontextprotocol/server-filesystem'
- ./examples
sortingKey: a3
type: agent
id: notebook-1
name: Sales Analysis
version: 1.0.0