{"version":"2.1.224","anchor":"assistant-text-is-scanned-for-harness-shaped-tags","canonical_anchor":"assistant-text-is-scanned-for-harness-shaped-tags","heading":"Assistant text is scanned for harness-shaped tags","tier":"internal","area":"Permissions","scope":null,"heads_up":null,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.224\/e\/assistant-text-is-scanned-for-harness-shaped-tags","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.224","markdown":"### Assistant text is scanned for harness-shaped tags\n\nClaude's own replies are scanned for text imitating internal control markup, and matches counted only.\n\n**What**\n\nMain-session assistant messages are now checked for text that looks like Claude Code's own internal control markup, including system reminders, harness envelopes and token-count, session-context, wake and role-turn shaped lines. Nothing is rewritten; the matches are only counted and reported.\n\n**Details**\n\n- Runs for main-session, non-error assistant messages only, with no flag involved.\n\n**Evidence**\n\n`tengu_assistant_output_harness_content`\n\n- Area: Permissions\n- Tier: Under the hood\n- Useful: 2\/5\n- Signal: 4\/5"}