Claude attribution

Adam discloses when Claude wrote something for him. Two parts to that:

1. A visible boundary - the drafted text is block quoted so a reader can see where Claude's words start and stop without reading carefully.
2. A varied header line - the disclosure is non-negotiable, but the wording should not be a rubber stamp. Repeating "I asked Claude about this and this was the result:" on every thread reads like a bot signature and stops carrying information. Write a fresh line each time.

Why the boundary matters, in a reviewer's own words (Gutenberg#81397): "I can't tell from your comment if your reply is the Claude generated response or your own take on this topic... I want to feel like we're collaborating together, and it will help me a lot if I can know when I'm chatting with you, or chatting with an LLM."

Pairs with adams-voice, which governs the writing other than the attribution. This skill only governs the attribution and the boundary.

Shape of an attributed comment

Three parts, in this order:

1. Adam's own line, outside the quote. One or two sentences in his voice - what he thinks, what he wants from the reader, what he plans to do next. This is what keeps the thread a conversation between people rather than a relay of machine output. Skip it only when the handoff really is all there is to say.
2. The attribution header, outside the quote. Italic, on its own line. Adam is speaking here, so "I" is correct.
3. The Claude-drafted body, block quoted. Every line prefixed with > , including blank lines between paragraphs, so the quote bar runs the full height of the response.