{"version":"2.1.295","anchor":"headless-compact-boundaries-and-transcript-rows-tracked-per","canonical_anchor":"headless-compact-boundaries-and-transcript-rows-tracked-per","heading":"Headless mode adds limits on schemas and on very large content lists","tier":"notice","area":"Headless Mode","scope":"individual","heads_up":true,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.295\/e\/headless-compact-boundaries-and-transcript-rows-tracked-per","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.295","markdown":"### Headless mode adds limits on schemas and on very large content lists\n\nIn headless mode, some sessions now refuse to compile schemas, and content lists over a size cap are rejected as too many items\n\n**Unclear.** It is not clear which sessions refuse to compile schemas.\n\n**What**\n\nHeadless mode is Claude Code running without its interactive screen, for example from a script. Several changes arrive there:\n\n- Some sessions now refuse to compile a JSON schema (a description of the shape some data must have), with the message \"this session compiles no schemas\". Before, schemas were always compiled.\n\n- A list of content pieces that is over a size cap is now rejected with the problem \"too_many_items\". Before, there was no cap.\n\n- Claude Code now tracks which user messages belong to each point where the conversation was compacted (shortened to save space).\n\n**Why**\n\nThese are new limits on schema and content handling, so very large inputs or schema use in some sessions can now fail where they did not before.\n\n- Area: Headless Mode\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: yes"}