FeedSummary.com Summary prompts

Syndication + normalization

Feed Formats for AI Summaries

RSS, Atom, JSON Feed, and Markdown can share one summarization pipeline when their original semantics are preserved during normalization.

RSSAtomJSON FeedMarkdowncontent blocks

Format map

Choose by source quality, not fashion

A robust consumer supports multiple formats and evaluates completeness, stability, identity, and update behavior per source.

RSS

Maximum source coverage, flexible conventions, and common full-content extensions.

RSS implementation

Markdown

A readable intermediate representation for normalized headings, lists, code, and links.

Markdown workflow

Data model

One internal record

Every parser should produce the same core fields while preserving format-specific values.

Feed formats converging into a normalized record

Normalize once. Summarize consistently.

  • Source-scoped item ID and canonical URL
  • Original and normalized title
  • Published and modified timestamps
  • Authors, tags, links, and attachments
  • Ordered content blocks and coverage status
  • Raw-field provenance and parse warnings

Normalization rules

Parser rules that prevent downstream errors

Preserve ambiguity instead of hiding it.

Identity and dates

Do not assume every RSS GUID is a URL, every Atom update is a publication date, or every JSON date is present. Keep original strings and the mapping decision.

Use source-scoped IDs, canonical URLs, and content fingerprints together.

Content and links

Retain the difference between summary and full content, text and HTML, alternate and self links, and article versus enclosure resources.

Convert meaningful content into blocks before chunking, and record when only an excerpt was available.

Feed Signals

Format comparison guide

A developer-focused comparison of RSS, Atom, and JSON Feed.