xonika9 Explain again
Repair understanding of one fragment of the immediately previous substantive assistant answer. This is a model-invoked skill in Claude Code and Codex; portable means structural compatibility only, not verified invocation behavior.
Trigger boundary
Use this skill only when the user's message is explicit confusion about the immediately preceding substantive answer in the same thread. That answer is an assistant response to the user, excluding tool output, status updates, system messages, and imported older chats.
Do not treat a general request to explain a topic as a reference to a prior answer. Explaining a provided article or source remains a source-explanation task. Creating a durable visual teaching artifact belongs to ce-explain; creating a visual scheme belongs to x9-diagrams.
Response
Identify the specific claim, conclusion, example, or step that is unclear. Add only the context needed to make that fragment understandable, then use a different explanatory method: an analogy, a concrete example, a causal breakdown, or simpler words. Include a tiny example only when it resolves the confusion.
Do not merely shorten or repeat the earlier wording. Do not create files, invoke ce-explain or x9-diagrams automatically, or start an autonomous loop. If the user still does not understand, one more bounded explanation is allowed; each new request still needs a clear referent.
When the unclear referent cannot be determined, ask one short question and do not invent context.
Completion
Return either a concise explanation that addresses the identified fragment through a different method, or one short clarifying question when the fragment is not identifiable.