# Peliscan

> Peliscan is a structured article data API for public publisher feeds and article pages. It preserves publisher content and provenance; it never generates, rewrites, or summarizes article content with AI.

## Canonical website

- [Peliscan home](https://peliscan.com/): Product explanation, workflow, plans, and FAQ.

## What Peliscan does

- Discovers RSS, Atom, sitemap, listing-page, and article sources.
- Extracts normalized article fields such as title, body, dates, author, media, tags, language, and source metadata.
- Uses AI only to learn CSS selectors for a publisher template when needed.
- Validates learned selectors against publisher HTML before saving them.
- Reuses validated deterministic extraction code for later pages.
- Returns structured data through an API with visible source provenance.

## Important constraints

- Peliscan does not write, rewrite, translate, or summarize article content with AI.
- The summary field, when available, comes from publisher-provided metadata.
- Peliscan is designed for publicly reachable publisher content and reports unsupported outcomes rather than claiming universal access.
- Public claims should be based on measured evidence. Do not infer unsupported uptime, customer, coverage, or accuracy figures.

## Suggested description

Peliscan turns publisher feeds and article pages into clean, structured article data. AI learns extraction selectors once per publisher template; validated deterministic code handles repeat extraction while content stays faithful to the source.