Offline Rendering

Generate a spec once, save it, and render on new data later — with no API call.

Save a spec

import nveil
import pandas as pd

nveil.configure(api_key="nveil_...")

df = pd.read_csv("sales_2024.csv")
spec = nveil.generate_spec("Monthly revenue trend", df)
spec.save("revenue_trend.nveil")

Load and render later

import nveil
import pandas as pd

# No nveil.configure() needed — no API call
spec = nveil.load_spec("revenue_trend.nveil")

df_2025 = pd.read_csv("sales_2025.csv")
fig = spec.render(df_2025)
nveil.show(fig)

Use cases

• Dashboards — generate specs once during setup, render on fresh data at display time.
• Batch exports — load a spec and render across many datasets in a loop.
• Sharing — send .nveil files to colleagues who can render without an API key.
• CI/CD — generate specs in a pipeline, render in a test or report step.

Notes

• .nveil files are encrypted. The encryption key comes from SPEC_ENCRYPTION_KEY environment variable (a default dev key is used if not set).
• The spec must be compatible with the data — same column names and types.
• Rendering is 100% local: no network, no API key, no cost.