diff --git a/scripts/aw.sh b/scripts/aw.sh
index da22e566c..0f6dbe166 100755
--- a/scripts/aw.sh
+++ b/scripts/aw.sh
@@ -90,8 +90,8 @@ for field in ("name", "trigger", "engine", "permissions"):
 
 # Validate engine — warn on unknown but don't fail (new models can be added without a code change)
 known_engines = {
-    "claude-sonnet-4-6", "claude-haiku-4-5-20251001", "claude-opus-4-7",
-    "claude-haiku-4-5",
+    "claude-sonnet-4-6", "claude-sonnet-5-0", "claude-haiku-4-5-20251001",
+    "claude-opus-4-7", "claude-haiku-4-5",
 }
 engine = fm.get("engine", "")
 if engine and engine not in known_engines:
diff --git a/scripts/engine.sh b/scripts/engine.sh
index f59d91706..292522efe 100644
--- a/scripts/engine.sh
+++ b/scripts/engine.sh
@@ -125,7 +125,7 @@ set_engine_config() {
       ENGINE_AUDIT_MODEL="claude-fable-5"
       ENGINE_ACTION_MODEL="claude-sonnet-4-6"
       ENGINE_SINGLE_MODEL="claude-fable-5"
-      ENGINE_LABEL="triage: haiku 4.5 → deep: opus 4.8 + duck: o4-mini → audit: fable 5"
+      ENGINE_LABEL="triage: haiku 4.5 [sonnet 4.6 → sonnet 5] → deep: opus 4.8 [sonnet 4.6 → sonnet 5] + duck: o4-mini → audit: fable 5"
       ENGINE_SINGLE_LABEL="single-reviewer mode: fable 5"
       # Cross-engine rubber duck: use Copilot when Claude is primary
       DUCK_ENGINE="copilot"
@@ -140,8 +140,16 @@ set_engine_config() {
       #   - adaptive thinking only; budget_tokens/temperature/top_p/top_k removed
       #   - omit thinking param entirely (disabled returns 400 on fable-5)
       #   - min cacheable prefix: fable-5 = 2048 tok, opus-4-8 = 4096 tok
-      CLAUDE_TRIAGE_MODEL_CHAIN="${CLAUDE_TRIAGE_MODEL_CHAIN:-claude-haiku-4-5-20251001,claude-sonnet-4-6}"
-      CLAUDE_DEEP_MODEL_CHAIN="${CLAUDE_DEEP_MODEL_CHAIN:-claude-opus-4-8,claude-sonnet-4-6}"
+      # Sonnet 5 candidate (#1098): claude-sonnet-5-0 is appended AFTER the
+      # sonnet-4-6 slot in the triage and deep chains as a non-default fallback
+      # candidate — it earns its own TPM/RPM bucket for the layer-1 rate-limit
+      # walk without displacing the current primaries (haiku triage / opus deep)
+      # or the sonnet-4-6 fallback ordering. It is NOT the fleet default; a
+      # promotion to primary rides Story 5's canary rings. The daily cap is
+      # shared across Claude models (#206), so this hop only helps per-model RPM/
+      # TPM, not the subscription cap. Action tier is out of scope (#1098).
+      CLAUDE_TRIAGE_MODEL_CHAIN="${CLAUDE_TRIAGE_MODEL_CHAIN:-claude-haiku-4-5-20251001,claude-sonnet-4-6,claude-sonnet-5-0}"
+      CLAUDE_DEEP_MODEL_CHAIN="${CLAUDE_DEEP_MODEL_CHAIN:-claude-opus-4-8,claude-sonnet-4-6,claude-sonnet-5-0}"
       CLAUDE_AUDIT_MODEL_CHAIN="${CLAUDE_AUDIT_MODEL_CHAIN:-claude-fable-5,claude-opus-4-8,claude-opus-4-7}"
       CLAUDE_ACTION_MODEL_CHAIN="${CLAUDE_ACTION_MODEL_CHAIN:-claude-sonnet-4-6,claude-opus-4-8}"
       CLAUDE_SINGLE_MODEL_CHAIN="${CLAUDE_SINGLE_MODEL_CHAIN:-claude-fable-5,claude-opus-4-8,claude-opus-4-7}"
diff --git a/tests/dev-lead/unit/test_engine_chain.bats b/tests/dev-lead/unit/test_engine_chain.bats
index d4976f973..eb9a2877c 100644
--- a/tests/dev-lead/unit/test_engine_chain.bats
+++ b/tests/dev-lead/unit/test_engine_chain.bats
@@ -185,6 +185,43 @@ _source_engine() {
   grep -q "claude-opus-4-8" "$MODEL_RECORD"
 }
 
+# ── Sonnet 5 candidate wiring (#1098) ─────────────────────────────────────────
+# Sonnet 5 is appended as a FALLBACK CANDIDATE after claude-sonnet-4-6 in the
+# triage and deep chains (not a replacement, not the fleet default). The action
+# tier is deliberately out of scope.
+
+@test "sonnet-5 candidate: triage chain default appends claude-sonnet-5-0 after sonnet-4-6" {
+  _source_engine "claude"
+  [ "$CLAUDE_TRIAGE_MODEL_CHAIN" = "claude-haiku-4-5-20251001,claude-sonnet-4-6,claude-sonnet-5-0" ]
+}
+
+@test "sonnet-5 candidate: deep chain default appends claude-sonnet-5-0 after sonnet-4-6" {
+  _source_engine "claude"
+  [ "$CLAUDE_DEEP_MODEL_CHAIN" = "claude-opus-4-8,claude-sonnet-4-6,claude-sonnet-5-0" ]
+}
+
+@test "sonnet-5 candidate: deep opus-4-8 + sonnet-4-6 both rate-limited → claude-sonnet-5-0 reached" {
+  _source_engine "claude"
+  # Both existing deep models throttle; the walk must reach the new candidate.
+  export STUB_ENGINE_EXIT_BY_MODEL="claude-opus-4-8=1|claude-sonnet-4-6=1|claude-sonnet-5-0=0"
+  export STUB_ENGINE_RESPONSE_BY_MODEL="claude-opus-4-8=429 too many requests|claude-sonnet-4-6=429 too many requests|claude-sonnet-5-0=sonnet 5 did the work"
+
+  run run_agentic "$TEST_PROMPT" "claude-opus-4-8" "deep"
+
+  [ "$status" -eq 0 ]
+  grep -q "claude-opus-4-8" "$MODEL_RECORD"
+  grep -q "claude-sonnet-4-6" "$MODEL_RECORD"
+  grep -q "claude-sonnet-5-0" "$MODEL_RECORD"
+  [[ "$output" == *"sonnet 5 did the work"* ]]
+}
+
+@test "sonnet-5 candidate: action chain default is NOT touched (scope guard)" {
+  _source_engine "claude"
+  # Action tier is out of scope for #1098 — must stay sonnet-4-6 → opus-4-8.
+  [ "$CLAUDE_ACTION_MODEL_CHAIN" = "claude-sonnet-4-6,claude-opus-4-8" ]
+  [[ "$CLAUDE_ACTION_MODEL_CHAIN" != *"claude-sonnet-5"* ]]
+}
+
 # ── Gemini/Copilot unchanged: no chain applied ────────────────────────────────
 
 @test "gemini: in-engine chain uses gemini model (not claude)" {