@@ -358,52 +358,20 @@ export class WindsurfClient {
           }
         }
 
-        // Stall detection — two flavors:
-        //   (a) "cold stall": 30s+ ACTIVE but never saw any text or tool
-        //       call → planner is deadlocked before even starting to
-        //       produce output. Rotate account, don't make the user wait.
-        //   (b) "warm stall": we already streamed some text, but it hasn't
-        //       grown for 15s while status is still non-IDLE → planner is
-        //       stuck in a tool round-trip or upstream throttle. Accept
-        //       what we have as a complete response rather than waiting
-        //       out the full 180s maxWait with the client hanging.
+        // Cold stall: 30s+ ACTIVE but never saw any text or tool call.
         const elapsed = Date.now() - startTime;
-        // Cap at maxWait (180s): long-context requests can legitimately take
-        // that long to emit the first token from Cascade. Was 90s which
-        // still tripped on very long prompts (issue #5).
-        const coldStallMs = Math.min(maxWait, 30_000 + Math.floor(inputChars / 1500) * 5_000);
+        const effectiveChars = inputChars + (toolPreamble?.length ?? 0);
+        const coldStallMs = Math.min(maxWait, 30_000 + Math.floor(effectiveChars / 1500) * 5_000);
         if (elapsed > coldStallMs && sawActive && !sawText && seenToolCallIds.size === 0) {
           log.warn(`Cascade cold stall: ${elapsed}ms active without any text or tool call (threshold=${coldStallMs}ms, inputChars=${inputChars}), bailing`);
           endReason = 'stall_cold';
           const err = new Error(`Cascade planner stalled — no output after ${Math.round(coldStallMs / 1000)}s`);
           err.isModelError = true;
           throw err;
         }
-        if (sawText && lastStatus !== 1 && (Date.now() - lastGrowthAt) > NO_GROWTH_STALL_MS) {
-          const diag = {
-            msSinceGrowth: Date.now() - lastGrowthAt,
-            textLen: totalYielded,
-            thinkingLen: totalThinking,
-            stepCount: yieldedByStep.size,
-            toolCalls: seenToolCallIds.size,
-            lastStatus,
-          };
-          // Short-reply stall → treat as error so handlers/chat.js retries on
-          // another account. A 50-char preamble is worse than no reply at all
-          // because the client accepts it as "successful" and shows it to the
-          // user. Retry only if we haven't streamed anything substantial yet
-          // (if we did, partial delivery + idle end is fine).
-          if (totalYielded < STALL_RETRY_MIN_TEXT) {
-            log.warn('Cascade warm stall (short, retrying on next account)', diag);
-            endReason = 'stall_warm_retry';
-            const err = new Error('Cascade planner stalled after preamble — no progress for 25s');
-            err.isModelError = true;
-            throw err;
-          }
-          log.warn('Cascade warm stall (accepting partial)', diag);
-          endReason = 'stall_warm';
-          break; // return what we have as a successful response
-        }
+
+        // NOTE: warm stall check moved AFTER step loop (below) so
+        // lastGrowthAt reflects data read in this poll, not the previous one.
 
         // Any trajectory change counts as forward progress. A new step, a new
         // tool call proposal, or thinking growth all reset the stall timer so
@@ -482,6 +450,22 @@ export class WindsurfClient {
           }
         }
 
+        // Warm stall: text stopped growing for 25s while planner is active.
+        // Placed AFTER the step loop so lastGrowthAt is current-poll fresh.
+        if (sawText && lastStatus !== 1 && (Date.now() - lastGrowthAt) > NO_GROWTH_STALL_MS) {
+          const diag = { msSinceGrowth: Date.now() - lastGrowthAt, textLen: totalYielded, thinkingLen: totalThinking, stepCount: yieldedByStep.size, toolCalls: seenToolCallIds.size, lastStatus };
+          if (totalYielded < STALL_RETRY_MIN_TEXT) {
+            log.warn('Cascade warm stall (short, retrying on next account)', diag);
+            endReason = 'stall_warm_retry';
+            const err = new Error('Cascade planner stalled after preamble — no progress for 25s');
+            err.isModelError = true;
+            throw err;
+          }
+          log.warn('Cascade warm stall (accepting partial)', diag);
+          endReason = 'stall_warm';
+          break;
+        }
+
         // Check status
         const statusProto = buildGetTrajectoryRequest(cascadeId);
         const statusResp = await grpcUnary(
@@ -506,7 +490,8 @@ export class WindsurfClient {
           idleCount++;
           // Require at least a little text OR a long idle streak before
           // accepting "done", so we don't race the first visible chunk.
-          const canBreak = sawText ? idleCount >= 2 : idleCount >= 4;
+          const growthSettled = (Date.now() - lastGrowthAt) > pollInterval * 3;
+          const canBreak = sawText ? (idleCount >= 3 && growthSettled) : idleCount >= 4;
           if (canBreak) {
             // Final sweep
             const finalResp = await grpcUnary(
 
@@ -17,7 +17,7 @@
  *     the same LS and the same account or the cascade_id is meaningless.
  *   - A checked-out entry is removed from the pool. Concurrent second request
  *     with the same fingerprint falls back to a fresh cascade.
- *   - TTL 10 min; LRU eviction at 500 entries.
+ *   - TTL 30 min; LRU eviction at 500 entries.
  */
 
 import { createHash } from 'crypto';
@@ -57,27 +57,31 @@ function canonicalise(messages) {
  * may restructure content arrays, add tool_use blocks, or modify text,
  * causing hash mismatches and 0% hit rate. (#24)
  */
+function systemPrefix(messages) {
+  return messages
+    .filter(m => m.role === 'system')
+    .map(m => typeof m.content === 'string' ? m.content : JSON.stringify(m.content ?? ''))
+    .join('\0');
+}
+
 export function fingerprintBefore(messages, modelKey = '') {
   if (!Array.isArray(messages) || messages.length < 2) return null;
   const users = messages.filter(m => m.role === 'user');
   if (users.length < 2) return null;
-  return sha256(modelKey + '\0' + JSON.stringify(canonicalise(users.slice(0, -1))));
+  return sha256(modelKey + '\0' + systemPrefix(messages) + '\0' + JSON.stringify(canonicalise(users.slice(0, -1))));
 }
 
-/**
- * Fingerprint for the full conversation after this turn completes.
- * Uses all user messages (including current). The *next* request's
- * `fingerprintBefore` will hash users[:-1] which equals this value.
- */
 export function fingerprintAfter(messages, modelKey = '') {
   const users = messages.filter(m => m.role === 'user');
   if (!users.length) return null;
-  return sha256(modelKey + '\0' + JSON.stringify(canonicalise(users)));
+  return sha256(modelKey + '\0' + systemPrefix(messages) + '\0' + JSON.stringify(canonicalise(users)));
 }
 
 function prune(now) {
+  for (const [fp, e] of _pool) {
+    if (now - e.lastAccess > POOL_TTL_MS) { _pool.delete(fp); stats.expired++; }
+  }
   if (_pool.size <= POOL_MAX) return;
-  // Drop oldest entries until back under the cap.
   const entries = [..._pool.entries()].sort((a, b) => a[1].lastAccess - b[1].lastAccess);
   const toDrop = entries.length - POOL_MAX;
   for (let i = 0; i < toDrop; i++) {
@@ -100,6 +104,7 @@ export function checkout(fingerprint) {
   _pool.delete(fingerprint);
   if (Date.now() - entry.lastAccess > POOL_TTL_MS) {
     stats.expired++;
+    stats.misses++;
     return null;
   }
   stats.hits++;
 
@@ -257,10 +257,7 @@ export async function handleChatCompletions(body) {
   // instead of replaying the whole history.
   //
   // Conversation reuse lets Cascade keep server-side context across turns.
-  // Previously disabled for tool-emulation but that caused ALL Claude Code
-  // conversations to lose context after every turn (#24). A fingerprint miss
-  // just falls back to fresh cascade (no worse than before). (#24)
-  const reuseEnabled = useCascade && isExperimentalEnabled('cascadeConversationReuse');
+  const reuseEnabled = useCascade && !emulateTools && isExperimentalEnabled('cascadeConversationReuse');
   const fpBefore = reuseEnabled ? fingerprintBefore(messages, modelKey) : null;
   let reuseEntry = reuseEnabled ? poolCheckout(fpBefore) : null;
   if (reuseEntry) log.info(`Chat: cascade reuse HIT cascadeId=${reuseEntry.cascadeId.slice(0, 8)}… model=${displayModel}`);
 
@@ -249,6 +249,11 @@ export class ToolCallStreamParser {
   }
 
   _consumeJsonBlock(parseFn, doneCalls, safeParts) {
+    if (this.buffer.length > 65_536) {
+      safeParts.push(this.buffer);
+      this.buffer = '';
+      return true;
+    }
     const endIdx = this._findClosingBrace();
     if (endIdx === -1) return false;
     const jsonStr = this.buffer.slice(0, endIdx + 1);
@@ -426,14 +431,24 @@ export class ToolCallStreamParser {
     }
     if (this.inToolCode) {
       this.inToolCode = false;
-      const tc = this._parseToolCodeJson(remaining);
-      if (tc) { this._totalSeen++; return { text: '', toolCalls: [tc] }; }
+      const endIdx = this._findClosingBrace();
+      if (endIdx !== -1) {
+        const jsonStr = remaining.slice(0, endIdx + 1);
+        const tail = remaining.slice(endIdx + 1);
+        const tc = this._parseToolCodeJson(jsonStr);
+        if (tc) { this._totalSeen++; return { text: tail, toolCalls: [tc] }; }
+      }
       return { text: remaining, toolCalls: [] };
     }
     if (this.inBareCall) {
       this.inBareCall = false;
-      const tc = this._parseBareToolCallJson(remaining);
-      if (tc) { this._totalSeen++; return { text: '', toolCalls: [tc] }; }
+      const endIdx = this._findClosingBrace();
+      if (endIdx !== -1) {
+        const jsonStr = remaining.slice(0, endIdx + 1);
+        const tail = remaining.slice(endIdx + 1);
+        const tc = this._parseBareToolCallJson(jsonStr);
+        if (tc) { this._totalSeen++; return { text: tail, toolCalls: [tc] }; }
+      }
       return { text: remaining, toolCalls: [] };
     }
     // Fallback: detect any remaining tool_code patterns in leftover buffer