{"version":"2.1.282","anchor":"forked-agents-get-a-never-consult-the-advisor-tool-messag","canonical_anchor":"forked-agents-get-a-never-consult-the-advisor-tool-messag","heading":"Forked agents are now told not to use the advisor tool","tier":"internal","area":"Advisor Tool","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.282\/e\/forked-agents-get-a-never-consult-the-advisor-tool-messag","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.282","markdown":"### Forked agents are now told not to use the advisor tool\n\nWhen the advisor is enabled, forked agents now get a one-time instruction never to consult the advisor tool, except for compaction forks\n\n**Unclear.** One further condition on whether the advisor counts as enabled was not resolved.\n\n**What**\n\nA forked agent is a helper agent that starts from a copy of the main conversation. The advisor is an experimental tool that lets Claude consult a second model. When the advisor is enabled, Claude Code now adds a hidden message saying \"Never consult the advisor tool.\" to each forked agent's conversation, once only. Forks made for compaction (shrinking the conversation to save space) are left out.\n\nThe advisor is enabled with the environment variable `CLAUDE_CODE_ENABLE_EXPERIMENTAL_ADVISOR_TOOL`, or by a remote switch that is off unless it is switched on remotely.\n\n**Why**\n\nThis stops forked agents from making extra advisor calls of their own.\n\n- Flag `tengu_sage_compass2`: Not enough to say (read for one account on one subscription tier against v2.1.282; this account: no value returned, anonymous baseline: no value returned, compiled default: not a boolean we can read) These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.\n- Area: Advisor Tool\n- Tier: Under the hood\n- Useful: 2\/5\n- Signal: 3\/5"}