Feat/llms txt generator

_config.yml

@@ -57,6 +57,34 @@ webmaster_verifications:
 # ↑ --------------------------
 # The end of `jekyll-seo-tag` settings
 
+# ------------ LLMs.txt settings › https://llmstxt.org/ ------------------
+# ↓ --------------------------
+
+llms_txt:
+  # Required
+  enabled: false     # false or omitted = plugin does nothing
+
+  # Optional — omit or leave empty to use defaults.
+  # llms_txt `enabled` must be 'true' for these to apply.
+  full:              # generate llms-full.txt (default: false)
+  title:             # default: site.title → "Site"
+  description: 
+
+  # Optional — extra links shown under "## Optional" section.
+  # Per the llmstxt.org spec, LLMs may skip this section when context is limited.
+  # Omit or leave empty to skip this section.
+
+  optional:
+    - name:            # e.g. RSS Feed
+      url:             # e.g. https://yourblog.com/feed.xml
+      description:     # e.g. Subscribe to new posts
+    - name:            # e.g. Resume
+      url:             # e.g. https://yourblog.com/resume.pdf
+      description:     # e.g. Professional background and experience
+
+# ↑ --------------------------
+# The end of LLMs.txt settings
+
 # Web Analytics Settings
 analytics:
   google:

_plugins/llms-txt-generator.rb

@@ -0,0 +1,245 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+#
+# Generate /llms.txt and optionally /llms-full.txt for LLM-friendly site content.
+# Spec: https://llmstxt.org/
+#
+# Contacts are auto-read from _data/contact.yml (no config needed).
+#
+# ── _config.yml ──────────────────────────────────────────────────────
+#
+#    llms_txt:
+#      # Required
+#      enabled: false                     # false or omitted = plugin does nothing
+#
+#      # Optional — omit or leave empty to use defaults.
+#      # llms_txt `enabled` must be 'true' for these to apply.
+#      full: true                        # generate llms-full.txt (default: false)
+#      title: "My Site"                  # default: site.title → "Site"
+#      description: "About my site"      # default: site.description → ""
+#
+#      # Optional — extra links shown under "## Optional" section.
+#      # Per the llmstxt.org spec, LLMs may skip this section when context is limited.
+#      # Omit or leave empty to skip this section.
+#      # optional:
+#      #   - name: RSS Feed
+#      #     url: https://yourblog.com/feed.xml
+#      #     description: Subscribe to new posts
+#      #   - name: GitHub
+#      #     url: https://github.com/yourname
+#      #     description: Open source projects and code
+#      #   - name: Resume
+#      #     url: https://yourblog.com/resume.pdf
+#      #     description: Professional background and experience
+#
+# ─────────────────────────────────────────────────────────────────────
+#
+# Minimal config (just these two lines):
+#
+#   llms_txt:
+#     enabled: true
+#
+
+require "nokogiri"
+
+module Jekyll
+  class LlmsTxtGenerator < Generator
+    safe true
+    priority :low
+
+    def generate(site)
+      return unless enabled?(site)
+
+      config = site.config["llms_txt"] || {}
+      posts = site.posts.docs.sort_by(&:date).reverse
+      pages = site.collections["tabs"]&.docs&.sort_by { |t| t.data["order"] || 0 } || []
+      contacts = resolve_contacts(site)
+
+      site.pages << build_llms_txt(site, config, posts, pages, contacts)
+      site.pages << build_llms_full_txt(site, config, posts, pages, contacts) if config["full"]
+    end
+
+    private
+
+    def enabled?(site)
+      site.config.dig("llms_txt", "enabled") == true
+    end
+
+    def site_url(site)
+      url = site.config["url"].to_s.chomp("/")
+      base = site.config["baseurl"].to_s.chomp("/")
+      "#{url}#{base}"
+    end
+
+    # ── Contact resolution ──────────────────────────────────────────────
+
+    # Reads _data/contact.yml and resolves each entry to a name + URL pair.
+    # Entries without a resolvable URL are silently skipped.
+    def resolve_contacts(site)
+      raw = site.data["contact"]
+      return [] if raw.nil? || !raw.is_a?(Array)
+
+      raw.each_with_object([]) do |entry, list|
+        type = entry["type"].to_s
+        url = entry["url"].to_s.strip
+        url = infer_contact_url(site, type) if url.empty?
+        next if url.empty?
+
+        url = "#{site_url(site)}#{url}" if url.start_with?("/")
+        list << { "name" => format_contact_name(type), "url" => url }
+      end
+    end
+
+    # Tries to build a URL from _config.yml values when contact.yml
+    # doesn't have an explicit url field.
+    def infer_contact_url(site, type)
+      case type
+      when "github"
+        username = site.config.dig("github", "username").to_s.strip
+        username.empty? ? find_social_link(site, "github.com") : "https://github.com/#{username}"
+      when "twitter"
+        username = site.config.dig("twitter", "username").to_s.strip
+        username.empty? ? find_social_link(site, "twitter.com") : "https://twitter.com/#{username}"
+      when "email"
+        email = site.config.dig("social", "email").to_s.strip
+        email.empty? ? "" : "mailto:#{email}"
+      when "rss"
+        "#{site_url(site)}/feed.xml"
+      else
+        find_social_link(site, type)
+      end
+    end
+
+    # Scans social.links array in _config.yml for a URL containing the keyword.
+    def find_social_link(site, keyword)
+      links = site.config.dig("social", "links")
+      return "" unless links.is_a?(Array)
+
+      links.find { |l| l.to_s.include?(keyword) }.to_s
+    end
+
+    # Turns a contact type slug into a readable name.
+    def format_contact_name(type)
+      type.split(/[-_]/).map(&:capitalize).join(" ")
+    end
+
+    # ── Section builders ────────────────────────────────────────────────
+
+    def build_llms_txt(site, config, posts, pages, contacts)
+      content = String.new
+      content << header(site, config)
+      content << pages_section(site, pages) unless pages.empty?
+      content << posts_section(site, posts) unless posts.empty?
+      content << contacts_section(contacts) unless contacts.empty?
+      content << optional_section(config)
+
+      make_page(site, "llms.txt", content)
+    end
+
+    def build_llms_full_txt(site, config, posts, pages, contacts)
+      content = String.new
+      content << header(site, config)
+
+      base = site_url(site)
+
+      unless pages.empty?
+        content << "## Pages\n\n"
+        pages.each do |page|
+          title = page.data["title"] || page.basename
+          url = "#{base}#{page.url}"
+          content << "### [#{title}](#{url})\n\n"
+          content << strip_html(page.content).strip << "\n\n"
+        end
+      end
+
+      unless posts.empty?
+        content << "## Posts\n\n"
+        posts.each do |post|
+          title = post.data["title"] || post.basename_without_ext
+          url = "#{base}#{post.url}"
+          desc = post.data["description"].to_s.strip
+          content << "### [#{title}](#{url})\n\n"
+          content << "> #{desc}\n\n" unless desc.empty?
+          content << strip_html(post.content).strip << "\n\n"
+        end
+      end
+
+      content << contacts_section(contacts) unless contacts.empty?
+      content << optional_section(config)
+
+      make_page(site, "llms-full.txt", content)
+    end
+
+    def header(site, config)
+      title = config["title"] || site.config["title"] || "Site"
+      desc = config["description"] || site.config["description"] || ""
+
+      out = String.new("# #{title}\n\n")
+      out << "> #{desc}\n\n" unless desc.empty?
+      out
+    end
+
+    def pages_section(site, pages)
+      base = site_url(site)
+      out = String.new("## Pages\n\n")
+      pages.each do |page|
+        title = page.data["title"] || page.basename
+        url = "#{base}#{page.url}"
+        out << "- [#{title}](#{url})\n"
+      end
+      out << "\n"
+    end
+
+    def posts_section(site, posts)
+      base = site_url(site)
+      out = String.new("## Posts\n\n")
+      posts.each do |post|
+        title = post.data["title"] || post.basename_without_ext
+        url = "#{base}#{post.url}"
+        desc = post.data["description"].to_s.strip
+        out << "- [#{title}](#{url})"
+        out << ": #{desc}" unless desc.empty?
+        out << "\n"
+      end
+      out << "\n"
+    end
+
+    def contacts_section(contacts)
+      out = String.new("## Contacts\n\n")
+      contacts.each do |c|
+        out << "- [#{c['name']}](#{c['url']})\n"
+      end
+      out << "\n"
+    end
+
+    def optional_section(config)
+      extras = config["optional"] || []
+      return "" if extras.empty?
+
+      out = String.new("## Optional\n\n")
+      extras.each do |item|
+        out << "- [#{item['name']}](#{item['url']})"
+        out << ": #{item['description']}" if item["description"]
+        out << "\n"
+      end
+      out << "\n"
+    end
+
+    def strip_html(text)
+      Nokogiri::HTML.fragment(text).text
+    end
+
+    def make_page(site, name, content)
+      page = PageWithoutAFile.new(site, site.source, "", name)
+      page.data = {
+        "layout" => nil,
+        "sitemap" => false,
+        "permalink" => "/#{name}"
+      }
+      page.content = content
+      page.output = content
+      page
+    end
+  end
+end