Supabase RSS Action

GitHub composite action that generates an RSS 2.0 feed (rss.xml) from any PostgREST/Supabase relation and writes it into your repository, so it can be committed or deployed with your static site.

Usage

jobs:
  generate-rss:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: AFischbeck/supabase-rss-action@v1
        with:
          supabase-url: ${{ secrets.SUPABASE_URL }}
          supabase-key: ${{ secrets.SUPABASE_KEY }}
          base-url: https://your-site.com
          output-dir: public
          table: posts
          filter: is_public=eq.true
          order: created_at.desc
          feed-title: Your Site
          feed-description: Latest posts from your-site.com
          feed-entries: '[{"path": "/post/{slug}", "title": "title", "pubdate": "created_at", "description": "body"}]'

Inputs

Input	Required	Default	Description
supabase-url	no*	—	Supabase project URL
supabase-key	no*	—	Supabase API key (anon or service role)
base-url	yes	—	Base URL of the site
table	no*	—	PostgREST relation (table or view) to fetch rows from; required when feed-entries is set
feed-title	yes	—	Channel <title>
feed-description	yes	—	Channel <description>
feed-link	no	base-url/	Channel <link>
feed-language	no	en	Channel <language>
feed-entries	no	[]	JSON array of row-route objects (see below); empty means a metadata-only feed and no API call is made
select	no	derived	Comma-separated columns to fetch; defaults to all columns referenced by feed-entries
filter	no	—	PostgREST predicate(s), e.g. is_public=eq.true. &-separated predicates are supported.
order	no	—	PostgREST order expression(s), e.g. created_at.desc
api-path	no	/rest/v1	PostgREST API path
limit	no	—	Maximum number of rows to fetch (PostgREST caps results, so set this if you have many rows)
output-dir	no	public	Directory to write rss.xml into

* table, supabase-url, and supabase-key are only needed when feed-entries is set; a metadata-only feed runs without them.

Feed entries

Each feed-entries entry produces RSS items from the fetched rows — one item per row, deduplicated by absolute link across entries:

{"path": "/post/{slug}", "title": "title", "pubdate": "created_at", "description": "body"}

Field	Required	Description
path	yes	URL template with {column} placeholders; used for both <link> and <guid> (guid = absolute link)
title	yes	Column name for <title>
pubdate	no	Column name (ISO timestamp) for <pubDate>, formatted RFC 822; omitted if unparseable
description	no	Column name (markdown) for <description>
description-length	no	Truncation length in chars (default 500)
strip-markdown	no	Strip markdown from the description (default true)

Rows missing title or any path placeholder column (or containing empty values) are skipped. <description> is always emitted, empty when the column is unset or the value is empty.

License

MIT