{"version":"2.1.302","anchor":"agent-summaries-gain-a-short-mode-that-summarizes-from-not","canonical_anchor":"agent-summaries-gain-a-short-mode-that-summarizes-from-not","heading":"Subagent progress summaries can work from short notes instead of the whole transcript","tier":"notice","area":"Subagents","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.302\/e\/agent-summaries-gain-a-short-mode-that-summarizes-from-not","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.302","markdown":"### Subagent progress summaries can work from short notes instead of the whole transcript\n\nProgress summaries for subagents can now be written from condensed notes, but only for opus or fable models when the summary setting is short\n\n**Unclear.** It is not known whether the short setting is on by default.\n\n**What**\n\nA subagent is a helper Claude starts to work on part of a task. While it runs, Claude Code writes short progress summaries of what it is doing in the background. Before, each summary was written by copying the subagent's whole transcript, its full record of messages, into a new request.\n\nThere is now a \"short\" mode. It turns the transcript into notes and asks for a summary from those notes instead. It is used only when both are true:\n\n- The summary settings ask for \"short\".\n\n- The subagent's model is from the opus or fable family.\n\nIn every other case the summary uses the full transcript as before. A summary can also now be skipped when there is nothing to describe, or no text to work from.\n\n**Why**\n\nWriting summaries from notes makes them cheaper when a subagent's transcript is large.\n\n- Area: Subagents\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: no"}