Atom Feed Generation

Eigen can auto-generate Atom 1.0 (RFC 4287) XML feeds for configured data collections. Feeds are written to dist/ at configurable paths during the build process.

Configuration

Add [feed.<name>] tables to site.toml. Each feed defines its data source (same fields as a frontmatter DataQuery) plus metadata and field mappings.

File-based example

[feed.blog]
file = "posts.json"        # file in _data/
sort = "-date"              # sort descending by date
limit = 20                  # max 20 entries
path = "blog/feed.xml"      # output path in dist/
link_prefix = "blog"        # entry URLs: {base_url}/blog/{slug}.html

Remote source example

[sources.cms]
url = "https://cms.example.com/api"

[feed.blog]
source = "cms"              # source from [sources.*]
query_path = "/posts"       # appended to source URL
root = "data.posts"         # dot-path into response JSON
sort = "-publishedAt"
title = "My Blog Feed"
author = "Jane Doe"
title_field = "title"
date_field = "publishedAt"
summary_field = "excerpt"
slug_field = "slug"
link_prefix = "blog"

Multiple feeds

[feed.blog]
file = "posts.json"
path = "blog/feed.xml"
link_prefix = "blog"

[feed.changelog]
file = "releases.json"
path = "changelog/feed.xml"
title_field = "version"
link_prefix = "changelog"

Configuration Reference

Field	Type	Default	Description
`file`	string	-	Local file in `_data/`. Required if `source` not set.
`source`	string	-	Source name from `[sources.*]`. Required if `file` not set.
`query_path`	string	-	URL path appended to source's base URL.
`root`	string	-	Dot-path into response JSON for array extraction.
`sort`	string	-	Sort spec: `"field"` ascending, `"-field"` descending.
`title`	string	`site.name`	Feed title.
`path`	string	`"feed.xml"`	Output path relative to `dist/`.
`author`	string	`site.schema.author`	Feed author name.
`limit`	usize	`50`	Maximum number of entries.
`title_field`	string	`"title"`	Field on each item for entry title.
`date_field`	string	`"date"`	Field on each item for entry date.
`summary_field`	string	-	Field on each item for entry summary. Omitted if unset.
`slug_field`	string	`"slug"`	Field on each item for URL slug.
`link_prefix`	string	-	URL prefix for entry links.

Validation

Each feed must have either file or source.
If source is set, it must exist in [sources.*].
path must not be empty.
limit must be > 0.
slug_field must not be empty.

Entry Link Construction

Entry URLs are built as: {base_url}/{link_prefix}/{slug}.html

If link_prefix is not set: {base_url}/{slug}.html

Date Handling

Supported input formats for the date field:

RFC 3339: 2026-03-17T12:00:00Z
Date only: 2026-03-17 (midnight UTC assumed)
With timezone: 2026-03-17T12:00:00+05:30

Dates are output as RFC 3339 in the Atom <updated> element. If a date cannot be parsed, the build timestamp is used as fallback.

Error Handling

Missing required fields (title, slug) on an entry: entry is skipped with a warning, not a build failure.
Invalid feed config: build error via validation.
Empty collections: valid feed with zero entries.

Architecture

Module: src/build/feed.rs
Pipeline position: after sitemap generation, before post-build hooks
Reuses DataFetcher (cache-warm from page rendering) and escape_xml from src/build/sitemap.rs
No external XML library; string-based XML construction
No new crate dependencies (uses existing chrono, serde, eyre)

Autodiscovery

Autodiscovery <link> tags are not injected automatically. Add this to your base template's <head> manually:

<link rel="alternate" type="application/atom+xml"
      title="My Blog Feed" href="/blog/feed.xml" />