Agent-Friendly Blog Implementation Plan

For agentic workers: REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (- [ ]) syntax for tracking.

Goal: Publish a minimal agent-friendly layer for the Jekyll blog with llms.txt, website skill discovery, a public SKILL.md, and lightweight article metadata improvements.

Architecture: Add three static public entry points at the site root and .well-known, then strengthen post metadata so the published content is easier for agents to identify and cite. Keep the first version static-only and route article submission through a documented email workflow.

Tech Stack: Jekyll, Minimal Mistakes, Liquid templates, static Markdown/JSON/TXT assets


Task 1: Publish static agent entry points

Files:

Write a concise machine-readable overview of the site, key pages, topics, and citation guidance.

Create /.well-known/skills/index.json that advertises the single sunxie-blog skill and points to the public SKILL.md.

Write skills/sunxie-blog/SKILL.md with Discover, Consume, and Submit sections plus the email submission contract.

Update _config.yml so .well-known is included in the generated site.

Task 2: Strengthen article metadata

Files:

Expose discoverable links such as llms.txt and the skill index from the shared head include.

Emit a compact BlogPosting JSON-LD block for posts so machines can identify canonical URL, author, dates, excerpt, tags, and language more reliably.

Task 3: Verify publication behavior

Files:

Run: bundle exec jekyll build

Confirm the build outputs:

Check that a built article page contains the new JSON-LD and discovery links.