PH Catalog / a local research cabinet

Find the product.
Read its evidence.

A resumable local Product Hunt catalogue for developers and product researchers. Import, verify and explore records with Python, DuckDB and Parquet — with source fields and unknowns kept explicit.

Three original fictional fixtures. No live market data or hosted app.

Retained source fieldsSeparate derived annotationsExplicit unknowns

01 / The local workflow

A catalogue you can trace.

Import & resume

The Python pipeline normalizes identities, follows canonical aliases and saves progress so interrupted imports can continue.

Verify & snapshot

Explore records in DuckDB and export Parquet snapshots. Keep verification and field provenance alongside the data.

Inspect the boundary

Atlas separates source fields from derived research annotations. Missing evidence stays unavailable; it is never filled with a guess.

02 / Original parser evidence

Three records. An honest starting point.

These fictional pages demonstrate the parser’s record shape and source fallbacks. They are neither live products nor an analytics mart.

03 / Your machine, your chosen mart

Run Atlas locally.

The operational Atlas UI is a read-only local analytics interface. It needs an already-built analytics mart; the three-record source catalogue alone is not that mart.

Follow the repository’s site/README.md and internal-analytics documentation to prepare the mart and open the loopback UI yourself.

With the analytics mart already built

uv run python tools/serve_analytics.py --database data/analytics/catalog-analytics.duckdb --port 8765

The public fixtures stay static. This page does not start collection or open a local service.

Read the source & documentation ↗