Authored result illustration · SHEET

Write product descriptions in bulk

A description for every product in your list, in one consistent voice.

Reuse this workflow

Start with this exact skill.

In a connected assistant, copy this public starter. Add your own facts in the conversation after retrieval. This starter contains no personal task inputs.

Open Claude ↗

Use LazyCoder get_skill to retrieve "product-descriptions-batch" at version "recipe-a765a5e54812ab51bbeff80d". Confirm release "publication-2cc5f3d871330e560bedbc0a" before using it; if unavailable, ask me to refresh the skill page. Follow its execution.conversation guide, instructions and output contract. Ask only missing questions in chat, at most three at a time, and wait for my answers before producing the result. Skip facts already provided; do not invent them. Keep my task details and result in this chat; do checks here, never with hosted validate_inputs or check_output.

Copy the starter, open Claude, and paste it into a chat with LazyCoder enabled. Opening Claude does not send your request or confirm a connection.

Prefer to prepare a filled prompt first? Customize the same skill. Connection trouble? Recover while keeping this skill selected.

From example to your own result

  1. Keep the structure. Use the required sections and checks below as a starting point.
  2. Replace the example facts. Bring Your product list. Missing facts should remain questions, not invented details.
  3. Review before use. Inspect the final document or file in the intended app. An illustration proves the preview renderer, not the assistant’s output quality.

What to check in your result

  • One row for every product, keyed by SKU, with a count that matches your list
  • Every description uses the same voice and structure, within the length you chose
  • Only facts from your list; missing details are flagged, not made up
  • Similar products and variants each get their own copy
  • A CSV or sheet ready to import, or to paste where the app can't make files
  • Long lists are done in batches of 25, with a running count

Structural checks do not establish factual accuracy or usefulness. How we review.