fix(#2852): forward model overrides from categories/agent config to subagents

- fix(switcher): use lastIndexOf for multi-slash model IDs (e.g. aws/anthropic/claude-sonnet-4)
- fix(model-resolution): same lastIndexOf fix in doctor parseProviderModel
- fix(call-omo-agent): resolve model from agent config and forward to both
  background and sync executors via DelegatedModelConfig
- fix(subagent-resolver): inherit category model/variant when agent uses
  category reference without explicit model override
- test: add model override forwarding tests for call-omo-agent
- test: add multi-slash model ID test for switcher

src/hooks/preemptive-compaction.test.ts

@@ -451,7 +451,9 @@ describe("preemptive-compaction", () => {
 
     expect(ctx.client.session.summarize).toHaveBeenCalledTimes(1)
 
-    // when - new message with high tokens (context grew after compaction)
+    // when - advance past the 60s cooldown window, then new message with high tokens
+    const originalNow = Date.now
+    Date.now = () => originalNow() + 61_000
     await hook.event({
       event: {
         type: "message.updated",
@@ -480,6 +482,7 @@ describe("preemptive-compaction", () => {
 
     // then - summarize should fire again
     expect(ctx.client.session.summarize).toHaveBeenCalledTimes(2)
+    Date.now = originalNow
   })
 
   // #given modelContextLimitsCache has model-specific limit (256k)

src/hooks/preemptive-compaction.ts

@@ -10,6 +10,7 @@ import { createPostCompactionDegradationMonitor } from "./preemptive-compaction-
 
 const PREEMPTIVE_COMPACTION_TIMEOUT_MS = 120_000
 const PREEMPTIVE_COMPACTION_THRESHOLD = 0.78
+const PREEMPTIVE_COMPACTION_COOLDOWN_MS = 60_000
 
 declare function setTimeout(handler: () => void, timeout?: number): unknown
 declare function clearTimeout(timeoutID: unknown): void
@@ -68,6 +69,7 @@ export function createPreemptiveCompactionHook(
 ) {
   const compactionInProgress = new Set<string>()
   const compactedSessions = new Set<string>()
+  const lastCompactionTime = new Map<string, number>()
   const tokenCache = new Map<string, CachedCompactionState>()
 
   const postCompactionMonitor = createPostCompactionDegradationMonitor({
@@ -85,6 +87,9 @@ export function createPreemptiveCompactionHook(
     const { sessionID } = input
     if (compactedSessions.has(sessionID) || compactionInProgress.has(sessionID)) return
 
+    const lastTime = lastCompactionTime.get(sessionID)
+    if (lastTime && Date.now() - lastTime < PREEMPTIVE_COMPACTION_COOLDOWN_MS) return
+
     const cached = tokenCache.get(sessionID)
     if (!cached) return
 
@@ -127,6 +132,7 @@ export function createPreemptiveCompactionHook(
       )
 
       compactedSessions.add(sessionID)
+      lastCompactionTime.set(sessionID, Date.now())
     } catch (error) {
       log("[preemptive-compaction] Compaction failed", { sessionID, error: String(error) })
     } finally {
@@ -142,6 +148,7 @@ export function createPreemptiveCompactionHook(
       if (sessionID) {
         compactionInProgress.delete(sessionID)
         compactedSessions.delete(sessionID)
+        lastCompactionTime.delete(sessionID)
         tokenCache.delete(sessionID)
         postCompactionMonitor.clear(sessionID)
       }

src/hooks/think-mode/switcher.test.ts

@@ -146,6 +146,19 @@ describe("think-mode switcher", () => {
         expect(getHighVariant("custom-llm/gemini-3.1-pro")).toBe("custom-llm/gemini-3-1-pro-high")
       })
 
+      it("should handle multi-slash model IDs (#2852)", () => {
+        // given model IDs with multiple slashes (e.g. aws/anthropic/claude-sonnet-4)
+        const variant = getHighVariant("aws/anthropic/claude-sonnet-4-6")
+
+        // then should split at last slash, preserving full provider prefix
+        expect(variant).toBe("aws/anthropic/claude-sonnet-4-6-high")
+      })
+
+      it("should return null for multi-slash unknown models", () => {
+        // given multi-slash model ID without high variant mapping
+        expect(getHighVariant("aws/anthropic/unknown-model")).toBeNull()
+      })
+
       it("should return null for prefixed models without high variant mapping", () => {
         // given prefixed model IDs without high variant mapping
         expect(getHighVariant("vertex_ai/unknown-model")).toBeNull()

src/hooks/think-mode/switcher.ts

@@ -26,9 +26,10 @@ import { normalizeModelID } from "../../shared"
  * extractModelPrefix("vertex_ai/claude-sonnet-4-6") // { prefix: "vertex_ai/", base: "claude-sonnet-4-6" }
  * extractModelPrefix("claude-sonnet-4-6") // { prefix: "", base: "claude-sonnet-4-6" }
  * extractModelPrefix("openai/gpt-5.4") // { prefix: "openai/", base: "gpt-5.4" }
+ * extractModelPrefix("aws/anthropic/claude-sonnet-4") // { prefix: "aws/anthropic/", base: "claude-sonnet-4" }
  */
 function extractModelPrefix(modelID: string): { prefix: string; base: string } {
-  const slashIndex = modelID.indexOf("/")
+  const slashIndex = modelID.lastIndexOf("/")
   if (slashIndex === -1) {
     return { prefix: "", base: modelID }
   }