How to share a Jupyter HTML report privately.

A published notebook snapshot should represent a known execution state. Restart and run the notebook as appropriate for your workflow, review warnings and failed cells, and make the data timestamp visible in the output.

• Confirm the notebook completed with the intended environment and inputs
• Label the reporting period and generation time
• Remove exploratory cells that do not belong in the shared report

Jupyter nbconvert includes an HTML exporter for turning an .ipynb notebook into a browser-viewable document. The default HTML export commonly embeds figures in the output, making it suitable for a single-file review workflow.

jupyter nbconvert --to html ./analysis.ipynb --output ./analysis-report.html

Notebook output can contain more than the final charts: credentials printed during debugging, query text, customer rows, local paths, stack traces, or hidden metadata. Inspect the exported HTML itself before publishing it.

• Clear cells that print tokens, headers, or connection strings
• Aggregate or redact personal and customer data
• Check hover text and embedded chart payloads, not only visible tables
• Open the exported file locally and search for known secret patterns

Private access reduces the audience; it does not sanitize the notebook or stop authorized viewers from inspecting its HTML source.

Use the CLI to publish the exported file. A slug such as analytics/weekly-retention gives the report a recognizable destination and can receive later versions without changing the link.

npx @internalpage/cli login

npx @internalpage/cli publish ./analysis-report.html --slug analytics/weekly-retention

Workspace access works for broadly shared operating metrics. Selected-email access is a better default for finance, customer-specific analysis, hiring exercises, security research, and reports under limited review.

• Use a separate page when a report needs a different audience
• Verify both an allowed and denied Google account
• Review access again when the report begins including new data

A scheduled job can execute the notebook, export HTML, validate that the file exists, and publish to the same slug. Fail closed when execution or data-quality checks fail so a broken or partial report does not replace the last good version.

• Keep the publishing token in the scheduler or CI secret store
• Do not publish when notebook execution fails
• Include freshness and source information in the report

npx @internalpage/cli publish ./analysis-report.html --slug analytics/weekly-retention --json