fix(multi-agent): recheck fanout and worker defaults, fix audit-defaults step 3

.changes/multi-agent/fix-multi-agent-defaults-haiku-5-5-recheck-3f9a1c7e.md

@@ -0,0 +1,8 @@
+---
+bump: patch
+---
+
+### Fixed
+
+- **`/multi-agent:audit-defaults` no longer tells the caller to pass `defaults` to `list-pointers.sh`.** The script reads that word as an owner and exits 2. Step 3 now says to pass only a role or `fanout`, or nothing.
+- **The `fanout` and `worker` defaults are rechecked as of 2026-10-10.** The cost page still recommends Opus 5.5 at medium effort, so no model or effort value changes. The two recheck triggers now say "a new model named on the cost page", because that page has no model table.

plugins/multi-agent/reference/defaults.yaml

@@ -24,8 +24,8 @@ fanout:
   model: opus
   pointer: https://code.claude.com/docs/en/workflows#cost
   pointer_cost: https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#orchestrator-strategy-delegate-bulk-work
-  as_of: 2026-10-02
-  recheck: a new model row on the cost page, or a workflows Cost section change to how a stage's model is chosen
+  as_of: 2026-10-10
+  recheck: a new model named on the cost page, or a workflows Cost section change to how a stage's model is chosen
 
 roles:
   orchestrator:
@@ -43,7 +43,7 @@ roles:
     pointer_cost: https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#orchestrator-strategy-delegate-bulk-work
     pointer_spawn: https://code.claude.com/docs/en/sub-agents#choose-an-effort-level
     as_of: 2026-10-10
-    recheck: a Claude Code release that edits the bundled /workflow-authoring skill or the workflows Cost section, a new model row on the cost page, the Agent tool's per-spawn effort parameter changes, or a role proposed for haiku (see routing-rubric.md)
+    recheck: a Claude Code release that edits the bundled /workflow-authoring skill or the workflows Cost section, a new model named on the cost page, the Agent tool's per-spawn effort parameter changes, or a role proposed for haiku (see routing-rubric.md)
   verifier:
     model: inherit
     effort: high

plugins/multi-agent/skills/audit-defaults/SKILL.md

@@ -53,8 +53,9 @@ web access.
 3. **Inputs.**
    - `defaults` mode: run
      `${CLAUDE_SKILL_DIR}/scripts/list-pointers.sh --json $ARGUMENTS` and keep
-     the array it prints as `pointers`. An unknown owner exits 2 and names the
-     valid ones.
+     the array it prints as `pointers`. Pass only an owner (`<role>` or
+     `fanout`) after `--json`; with none, pass nothing, never the mode name
+     `defaults`. An unknown owner exits 2 and names the valid ones.
    - `repo` mode: run `${CLAUDE_SKILL_DIR}/scripts/list-targets.sh` and keep
      the array it prints as `targets`. It lists tracked markdown that states a
      model, effort, workflow or subagent claim, grouped into areas, and leaves
@@ -138,8 +139,8 @@ Shows the resolved map after an accepted change lands.
 
 - A `current` verdict needs the fetched text. A page that fails to load is
   `unread`, never `current`.
-- Treat a new model row on the cost page as a recheck event for the worker
-  and the fan-out guard, even when nothing about the older models changed.
+- Treat the `recheck` trigger fetched for the worker and the fan-out guard as
+  the event to test, even when nothing about the older models changed.
 - The fan-out guard is a user requirement as well as an upstream-backed
   default: propose changing `fanout.model` when the evidence supports it, but
   never propose turning the guard off.