rss-feed-generator

name: rss-feed-generator description: > Generate valid RSS 2.0 or Atom 1.0 feeds from web pages that contain post lists but lack a native feed. Triggers on phrases like "generate a feed for", "create an RSS feed from", "make an Atom feed for", "this page has no RSS", or any request to produce a feed URL or feed XML from a blog/news/post-list page. The skill fetches the page, extracts post metadata, and outputs well-formed feed XML plus a self-hostable HTML wrapper with an embedded feed discovery link tag. license: See LICENSE.txt

RSS / Atom Feed Generator (v1.0.0)

Generates valid RSS 2.0 or Atom 1.0 XML feeds from any web page that lists posts, articles, or news items but does not publish its own feed.

Defaults & Settings

Feed Generation Harness Mode

Use Feed Generation Harness Mode whenever the user asks to generate, refresh, repair, or self-host an RSS/Atom feed from a web page.

Harness mode constrains interpretation to feed production. Do not drift into a generic page summary or link list when a feed artifact is requested.

Harness Contract

When active:

1. Template match first — match the request to the predefined feed trigger templates before any fetch or generation.
2. Discover before generating — check for existing RSS/Atom feeds via HTML <link rel="alternate">, HTTP Link headers where available, and common feed paths before synthesizing a new feed.
3. Extract canonical post items — collect absolute post URLs, titles, dates when available, authors when available, and stable descriptions/excerpts.
4. Generate valid XML — use the RSS 2.0 or Atom 1.0 templates, with absolute links, canonical GUID/ID values, valid dates, and UTF-8 encoding.
5. HTML wrapper when requested — include a self-hostable wrapper with embedded feed discovery metadata and clear download/open links.
6. Validation gate — run the validation checklist from references/validation-checklist.md; do not deliver until XML structure, required channel/feed fields, item IDs, dates, and absolute links pass.
7. Provenance — state source page URL, generation time, item count, format, and whether the feed was discovered or synthesized.

Predefined Prompt Templates

Always match the user's input to a trigger template before executing or falling back to general LLM knowledge. Substitute {url} with the URL provided by the user.

If no trigger phrase matches, fall through in this order: direct native execution → REST function execution → Terminal-owned OAuth flow → MCP → authenticated chatPromptComplete → OPAL Agent routing.

Execution Routing

Default execution order for REST/OpenAPI interactions:

1. WEB_FETCH and other functions via URIBurner REST
2. Terminal-owned OAuth flow — when the endpoint requires OAuth 2.0 authentication, execute the OAuth flow from the terminal (authorization code, client credentials, or device flow), capture the Bearer token, and inject it into subsequent REST/OpenAPI calls via Authorization: Bearer {token} headers
3. MCP via https://linkeddata.uriburner.com/chat/mcp/messages or https://linkeddata.uriburner.com/chat/mcp/sse
4. Authenticated LLM-mediated execution via https://linkeddata.uriburner.com/chat/functions/chatPromptComplete
5. OPAL Agent routing using recognizable OPAL function names
6. Direct curl or wget as last resort

If the user's prompt expresses a protocol preference, follow that preference instead of the default order. See references/protocol-routing.md for exact endpoint patterns.

Order of Operations

⛔ PRE-BUILD CHECK: Before producing output, re-read the relevant workflow section above and re-read any checklists or verification gates defined in this skill. Confirm each checklist item before writing output. Build to pass — do not retro-fit. Apply the CLAUDE.md Anti-Drift Protocol: re-read spec section before build, gate-first validation, section-by-section delivery.

1. Page Fetch — Retrieve the target URL by invoking the WEB_FETCH function (defined in https://linkeddata.uriburner.com/chat/functions/openapi.yaml, endpoint /WEB_FETCH). Use whichever available protocol applies — REST, MCP, OPAL, or curl. Required parameter: url. Optional: headers, max_redirects, timeout_seconds. WEB_FETCH retrieves the page just like a web browser and returns the full page content for subsequent processing. See protocol-routing.md for exact invocation patterns per protocol.
2. Feed Discovery — Check <link rel="alternate"> tags. If a feed already exists, report it and stop (offer to proxy/mirror instead).
3. Post Extraction — Use the heuristics in references/extraction-rules.md to identify post entries (title, URL, date, author, summary).
4. Feed Assembly — Build XML using the templates in references/feed-templates.md. Validate structure against the checklist in references/validation-checklist.md.
5. Output — Present the feed XML in a code block; optionally wrap in the HTML discovery page template from references/html-wrapper-template.md.
6. Download Link — Save the .xml file to /mnt/user-data/outputs/ and present it with present_files.

Extraction Heuristics (summary — full rules in references/)

Structural signals to look for

• Repeated <article>, <li>, or <div> blocks with consistent class names (e.g., post, entry, article, blog-post, card)
• <h1>/<h2>/<h3> headings inside each block → item title
• <a href> on or near the heading → item link (resolve relative URLs)
• <time datetime="…"> or text matching date patterns → pubDate
• First <p> inside the block (≤280 chars) → description / summary
• <span class="author"> or byline text → author

Date handling

URL normalisation

All item <link> values must be absolute. Resolve against the page's <base href> if present, otherwise against the origin of the page URL.

Feed Format Specs

RSS 2.0 required elements

<rss version="2.0">
  <channel>
    <title>, <link>, <description>   ← required channel fields
    <item>
      <title>, <link>, <guid>        ← required per item
      <pubDate>                      ← RFC 822 (e.g. Mon, 01 Jan 2024 00:00:00 +0000)
      <description>                  ← plain text or CDATA-wrapped HTML

Atom 1.0 required elements

<feed xmlns="http://www.w3.org/2005/Atom">
  <title>, <id>, <updated>           ← required feed fields
  <entry>
    <title>, <id>, <updated>         ← required per entry
    <link href="…" rel="alternate"/> ← post URL
    <summary> or <content>           ← excerpt or full body

Full templates → references/feed-templates.md

Output Modes

Default: file (saves XML and presents download link).

Error Handling

Commands

Operational Rules

1. Always check for an existing feed before generating a synthetic one.
2. Never fabricate post content — only use text found on the page.
3. Escape &, <, > in text nodes; use CDATA for HTML description bodies.
4. All <guid> / Atom <id> values must be the canonical post URL.
5. Sort items newest-first by default (pubDate DESC).
6. Do not request or store credentials; only scrape publicly accessible pages.
7. Clearly label generated feeds as synthetic (add <generator> tag).
8. Respect robots.txt — do not scrape pages that disallow crawlers.