SkillsAdd

smithery/deepinthedata

advertorial-swiper

Swipe/record advertorial and landing pages into durable artifacts: extract markdown via r.jina.ai/URL, take an above-the-fold screenshot (no scrolling), upload artifacts to Cloudflare R2, and append the result to a Notion page.

Installation

$ npx skills add smithery/deepinthedata --skill advertorial-swiper
View on skills.sh ↗ Website ↗

Summary

  • Swipe/record advertorial and landing pages into durable artifacts: extract markdown via r.jina.ai/URL, take an above-the-fold screenshot (no scrolling), upload artifacts to Cloudflare R2, and append the result to a Notion page.
  • Use when a user provides an advertorial URL and wants it archived for a swipe file with screenshot + extracted markdown.
More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,889 B
  • docs SUMMARY.md 374 B

History

  1. First recorded snapshot · 0 installs

SKILL.md

Advertorial Swiper

Swipe an advertorial end-to-end into durable artifacts.

What it does:

  • Loads the advertorial URL.
  • Scrolls to the end of the page.
  • Clicks the primary CTA (e.g. “Check availability now”, “Special offer”, etc.).
  • Continues through sales/checkout steps until it detects a billing/payment form.
  • For each page/step, captures:

- Step 1 (advertorial): step-01.jpg above-the-fold screenshot (no scrolling) - Steps 2+ (offer/checkout): step-02.jpg, step-03.jpg, ... as full-page screenshots - waits a couple seconds before screenshotting (configurable) - step-XX.html (best-effort) - step-XX.json metadata (incl. billing-form detection) - step-XX.md markdown via https://r.jina.ai/<URL> (default on)

Then sync artifacts to R2 (public URLs) and append the results to a Notion page.

Setup (first time only)

From the skill folder:

  • npm install
  • npx playwright install chromium

Workflow

1) Swipe the advertorial (multi-page)

Run:

  • node scripts/swipe.js "<advertorial_url>"

Optional env:

  • OUT_DIR=/path/to/outputs (default ./outputs)
  • RUN_ID=... (otherwise uses timestamp)
  • SWIPE_KEY=... (stable key; otherwise derived from URL)
  • MAX_PAGES=8
  • STOPATBILLING=true|false (default true)
  • VIEWPORTW=1280 / VIEWPORTH=720
  • JPG_QUALITY=82
  • SCROLL_SCREENSHOTS=true|false (default true)
  • WAITBEFORESCREENSHOT_MS=2000
  • MAXSCREENSHOTSPER_PAGE=30
  • CAPTURE_MARKDOWN=true|false (default false)

Output directory:

  • outputs/run-<swipeKey>-<runId>/...

2) Upload artifacts to R2 (deterministic)

Run:

  • SRCDIR=<outDir> node scripts/r2sync.js

Required env (recommend skill-local ./.env):

  • R2_ENDPOINT
  • R2_BUCKET
  • R2PUBLICBASE (e.g. https://pub-....r2.dev)
  • R2ACCESSKEY_ID
  • R2SECRETACCESS_KEY

Optional env:

  • R2_PREFIX (default page)

Notes:

  • r2_sync.js uploads only images (JPG) to R2 to save costs.

Writes:

  • r2-manifest.json into SRC_DIR

3) Append to Notion page

Run:

  • SRCDIR=<outDir> node scripts/publishnotion.js

Required env (in ./.env):

  • NOTIONAPIKEY
  • NOTIONPARENTPAGE_ID (preferred)

Optional env:

  • NOTIONPAGEID (fallback: append to an existing page)

Notes:

  • The markdown is appended as one or more Notion code blocks (language=markdown) to avoid “markdown→Notion” conversion complexity.
  • The screenshot is added as an external image (best when r2-manifest.json is present).