{"version":"2.1.286","anchor":"prompt-dump-state-reset-per-agent-and-retired-init-hashes","canonical_anchor":"prompt-dump-state-reset-per-agent-and-retired-init-hashes","heading":"Prompt dumps no longer rewrite an agent's starting data, and reset clears them","tier":"notice","area":"Elsewhere","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.286\/e\/prompt-dump-state-reset-per-agent-and-retired-init-hashes","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.286","markdown":"### Prompt dumps no longer rewrite an agent's starting data, and reset clears them\n\nPrompt dumping now avoids rewriting an agent's starting data a second time, and resetting now clears prompt-dump state\n\n**Unclear.** Which setting turns on prompt dumping is not known.\n\n**What**\n\nPrompt dumping is a debugging feature that writes out what Claude Code sends to the model, tracked separately for each agent (a separate Claude instance handling part of a task). Two things are different:\n\n- Once an agent's starting data has been written, Claude Code now marks it as done. A later dump for the same agent does not write that starting data again.\n\n- The reset step now clears the prompt-dump state. Before, it cleared request replay chains instead.\n\n**Why**\n\nThis changes how repeated debug output is avoided when prompts are dumped. It affects developers using this debugging output rather than normal use.\n\n- Area: Elsewhere\n- Tier: You'll notice\n- Useful: 1\/5\n- Signal: 1\/5"}