• Who We Are
• Blog

Contact Us

Portfolio sync

Weekly automated detection of new and updated entries on portfolio.apexure.com, writing skeleton drafts into _landing-page-examples/ for editorial review.

How it runs

• Cron: every Monday at 08:00 UTC, via .github/workflows/portfolio-sync.yml
• Manual: Actions tab → “Portfolio sync” → “Run workflow” (optional max_drafts input)
• Locally: ruby script/portfolio-sync/sync.rb (writes drafts to working tree)

What it does

1. Pulls every post from https://portfolio.apexure.com/wp-json/wp/v2/posts
2. Diffs against state.json — a manifest of every WP post ID we’ve seen and its last-known modified timestamp
3. For each truly-new or modified item, writes _landing-page-examples/<slug>-landing-page.md with:
   • Pre-filled metadata: title, featured image, WP categories → tags, date, portfolio-id and portfolio-url for re-matching, published: false
   • TODO markers on every editorial field (description, meta-keywords, industry, page-type, technology, audience, cro-score, primary-cta, conversion-goal, design-features, psychological-principles, FAQ section)
4. Updates state.json so subsequent runs only flag genuine deltas
5. Opens a PR titled Portfolio sync — new portfolio.apexure.com items
6. Posts to Slack (if SLACK_WEBHOOK_URL secret is configured)

Drafts are published: false, so Jekyll skips them at build time — they never appear in the lookbook listing or get a public URL until a human flips the flag.

Handoff to the “full implementor” agent

When you git checkout a sync branch in VSCode and ask Claude to fill in the drafts, this is the contract it implements per file:

The <!-- DRAFT auto-generated --> block at the bottom of each file is a checklist for the agent. Delete it once the file is fully authored.

CRO scoring rubric

Score 0-20 in each of five dimensions, sum to a total out of 100:

A strong B2B SaaS demo page typically scores 80-95. A page with distracting navigation, generic stock proof, and a fuzzy headline scores 50-70.

State file format

state.json:

{
  "last_run": "2026-05-02T11:02:37Z",
  "items": {
    "2823": {
      "slug": "valley-solar",
      "modified": "2026-04-28T08:16:16",
      "link": "https://portfolio.apexure.com/valley-solar/"
    }
  }
}

Keyed by WP post ID, so slug renames don’t break tracking. Committed to the repo so the cron is stateless across runs.

Backlog handling

When this system was set up, portfolio.apexure.com had 495 posts vs. 311 examples in _landing-page-examples/ — i.e. ~184 portfolio items have never been written up here. The cron does not auto-draft this backlog: it was seeded into state.json as “already known” so weekly runs only flag genuinely new items.

If you want to chip away at the backlog:

# Drop a slug from state.json and re-run sync — that one item gets drafted
ruby -rjson -e '
  s = JSON.parse(File.read("script/portfolio-sync/state.json"))
  s["items"].delete("2823")  # ← whichever WP post id
  File.write("script/portfolio-sync/state.json", JSON.pretty_generate(s) + "\n")
'
ruby script/portfolio-sync/sync.rb --max-drafts 1

Or batch through them with --max-drafts 5 once a week.

Slack setup

To enable the Slack notification:

1. Slack: create an app at https://api.slack.com/apps (or use an existing one)
2. App → “Incoming Webhooks” → activate → “Add New Webhook to Workspace” → pick the channel
3. Copy the webhook URL (starts with https://hooks.slack.com/services/...)
4. GitHub: repo → Settings → Secrets and variables → Actions → New repository secret
   • Name: SLACK_WEBHOOK_URL
   • Value: the URL from step 3

The workflow gracefully no-ops the Slack step if the secret isn’t set.

Operations