@alexeiled/pi-model-router 0.6.1 → 0.6.2

package/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## 0.6.2 - 2026-09-21
+
+- Increase the default Jev timeout from 750 ms to 1500 ms. User-level `jev.timeoutMs` now sets the total advisory budget without the previous hidden 750 ms cap or an arbitrary upper cap. Positive finite values within Node's timer range are accepted, including 4000 and 5000 ms. Existing explicit shorter timeouts remain valid.
+- Fix delayed Pi thinking-display events being mistaken for user changes and overriding every routing tier. Internal display updates now preserve configured per-tier effort; explicit user overrides still apply.
+- Keep privacy opt-in, confidence validation, caller cancellation, and direct baseline fallback unchanged. No additional advisory requests or retries.
+- Add regression coverage for delayed valid Jev responses, configurable deadlines, internal thinking events, and Astra/Luna candidate selection. Align configuration examples and operator documentation.
+
 ## 0.6.1 — 2026-09-21
 
 - Add compact human-readable route provenance to the Pi footer, widget and `/router status`: `🧭 Jev ✓` means Jev selected the route, `🧭 Jev ↪ base` means Jev ran but the local baseline was used, and no marker means Jev was not involved.

package/README.md

@@ -95,7 +95,7 @@ pi -e ./extensions/index.ts
 
 - Generation and classification use Pi's provider registry, including native/custom providers and credential-specific URLs. Only the optional Jev advisor uses separate HTTPS transport.
 - Fallbacks run only before content is emitted; cancellation does not retry. Every target must support the requested input and exact thinking level; explicit unsupported effort is not silently reduced. Omitted thinking defaults to `off` for non-reasoning targets, including fallbacks.
-- Jev gets at most 750 ms (or its shorter configured timeout and remaining time in the 1500 ms advisory budget), with no retry. The separate classifier-only compatibility path retains its 10-second bound and 256-token output limit. Failure or uncertainty means eligible baseline; caller cancellation stops generation.
+- Jev gets a configurable total advisory budget via `jev.timeoutMs`: 1500 ms by default, with no additional routing cap or retry. The separate classifier-only compatibility path retains its 10-second bound and 256-token output limit. Failure or uncertainty means eligible baseline; caller cancellation stops generation.
 - Valid same-turn tool continuations reuse the actual prior route before either advisor. Pins, budget policy and a single eligible primary candidate also bypass advisors. Invalid continuations choose a compatible local route without advice; incompatible Google thought-signature replay fails plainly.
 - Pi owns tool execution permissions and per-request authentication. The router checks configured provider/profile identity, not which backend login is currently behind a provider. No private authentication storage is read.
 - Context trimming preserves system instructions and whole active tool turns. It is a text estimate, not a guarantee that images or a large active turn fit.
@@ -199,7 +199,7 @@ profile opt-ins, are ignored with a warning, before merging user credentials.
     "apiKey": "<rendered by chezmoi/1Password>",
     "endpoint": "https://api.typesafe.ai/v1/systemone",
     "model": "jev-1.13.0",
-    "timeoutMs": 750,
+    "timeoutMs": 1500,
     "confidenceThreshold": 0.65,
     "maxStateChars": 12000,
     "mode": "advisory"
@@ -218,14 +218,22 @@ profile opt-ins, are ignored with a warning, before merging user credentials.
 
 The endpoint, model, timeout, confidence threshold, state limit and mode shown
 above are defaults. Only HTTPS endpoints without embedded credentials, query
-parameters or fragments are accepted. Timeout must be positive and at most
-1500 ms, confidence must be 0–1, and the context limit must be 1–12000 characters.
-Provider routing further caps Jev at 750 ms within the fixed 1500 ms advisory
-budget; increasing `timeoutMs` does not extend those caps. Values above 750 ms
-are normalized to 750 ms with a configuration warning. The separate classifier-only
+parameters or fragments are accepted. `timeoutMs` defaults to 1500 ms and must
+be a positive finite number within Node's timer range (at most 2147483647 ms).
+There is no product-level cap: 4000 or 5000 ms are valid if you prefer waiting
+longer before falling back. It sets the total Jev advisory budget, including
+request and response-body time; there is no separate 750 ms cap. Confidence must
+be 0–1, and the context limit must be 1–12000 characters. The separate classifier-only
 path keeps a 10-second bound. Neither path retries or starts generation after
 caller cancellation.
 
+If Jev frequently falls back because requests time out, try `"timeoutMs": 3000`
+in your user config. Existing explicit values such as 750 remain unchanged;
+remove the field or set it to 1500 to use the new default. Increasing the timeout
+does not lower the confidence threshold or guarantee a different route. After
+upgrading, start a new Pi session; use `/router thinking auto` to clear any
+unwanted effort override in an existing session.
+
 **External data:** Jev receives bounded, role-labelled recent user/assistant/tool
 text, prioritizing the latest user request within `maxStateChars`, plus candidate
 tier/model/thinking identifiers. Truncation is deterministic, with no keyword

package/extensions/config.ts

@@ -376,10 +376,13 @@ export const normalizeTierConfig = (
   };
 };
 
+// Node turns larger setTimeout delays into 1 ms rather than waiting longer.
+const MAX_TIMER_DELAY_MS = 2_147_483_647;
+
 export const DEFAULT_JEV_CONFIG = {
   endpoint: 'https://api.typesafe.ai/v1/systemone',
   model: 'jev-1.13.0',
-  timeoutMs: 750,
+  timeoutMs: 1500,
   confidenceThreshold: 0.65,
   maxStateChars: 12000,
   mode: 'advisory',
@@ -420,7 +423,7 @@ export const normalizeJevConfig = (
     typeof value.timeoutMs !== 'number' ||
     !Number.isFinite(value.timeoutMs) ||
     value.timeoutMs <= 0 ||
-    value.timeoutMs > 1500 ||
+    value.timeoutMs > MAX_TIMER_DELAY_MS ||
     typeof value.confidenceThreshold !== 'number' ||
     !Number.isFinite(value.confidenceThreshold) ||
     value.confidenceThreshold < 0 ||
@@ -434,10 +437,6 @@ export const normalizeJevConfig = (
       (typeof value.apiKey !== 'string' || /[\r\n]/.test(value.apiKey)))
   )
     return invalid();
-  if (value.timeoutMs > 750)
-    warnings.push(
-      'Jev timeoutMs clamped to the effective 750 ms provider cap.',
-    );
   const apiKey = typeof value.apiKey === 'string' ? value.apiKey.trim() : '';
   if (value.enabled === true && !apiKey) {
     warnings.push('Jev disabled: missing user-config API key.');
@@ -447,7 +446,7 @@ export const normalizeJevConfig = (
     apiKey,
     endpoint: value.endpoint,
     model: value.model,
-    timeoutMs: Math.min(value.timeoutMs, 750),
+    timeoutMs: value.timeoutMs,
     confidenceThreshold: value.confidenceThreshold,
     maxStateChars: value.maxStateChars,
     mode: 'advisory',

package/extensions/index.ts

@@ -57,6 +57,10 @@ const routerExtension = (pi: ExtensionAPI) => {
   let isInitialized = false;
   let isInternalModelSwitch = false;
   let isInternalThinkingChange = false;
+  const pendingThinkingChanges: Array<{
+    previousLevel: ThinkingLevel;
+    level: ThinkingLevel;
+  }> = [];
   let ignoreStartupThinkingEvent = false;
 
   const runtimeState = {
@@ -142,13 +146,27 @@ const routerExtension = (pi: ExtensionAPI) => {
   };
 
   const setThinkingLevelInternally = (level: ThinkingLevel) => {
+    let previousLevel: ThinkingLevel;
+    try {
+      previousLevel = pi.getThinkingLevel();
+    } catch {
+      return; // The session runtime may already be torn down.
+    }
+    const change = { previousLevel, level };
+    pendingThinkingChanges.push(change);
     isInternalThinkingChange = true;
     try {
       pi.setThinkingLevel(level);
+      change.level = pi.getThinkingLevel();
     } catch {
       // Extension context may be stale after session teardown.
+      change.level = change.previousLevel;
     } finally {
       isInternalThinkingChange = false;
+      if (change.level === change.previousLevel) {
+        const index = pendingThinkingChanges.indexOf(change);
+        if (index >= 0) pendingThinkingChanges.splice(index, 1);
+      }
     }
   };
 
@@ -513,8 +531,21 @@ const routerExtension = (pi: ExtensionAPI) => {
 
   pi.on('thinking_level_select', (event, ctx) => {
     ensureInitializedFromContext(ctx);
+    if (isInternalThinkingChange) {
+      pendingThinkingChanges.pop();
+      return;
+    }
+    // Pi emits without awaiting extension handlers; earlier handlers can delay this echo.
+    const internalChange = pendingThinkingChanges.findIndex(
+      (change) =>
+        change.previousLevel === event.previousLevel &&
+        change.level === event.level,
+    );
+    if (internalChange >= 0) {
+      pendingThinkingChanges.splice(internalChange, 1);
+      return;
+    }
     if (!isInitialized || !routerEnabled || !selectedProfile) return;
-    if (isInternalThinkingChange) return;
     if (ignoreStartupThinkingEvent) {
       ignoreStartupThinkingEvent = false;
       return;

package/extensions/provider.ts

@@ -486,7 +486,8 @@ export const registerRouterProvider = (
             advisorConfigured
           ) {
             const started = performance.now();
-            const routingDeadline = started + (useJev ? 1500 : 10_000);
+            const routingDeadline =
+              started + (useJev && jev ? jev.timeoutMs : 10_000);
             const candidates = primaryRoutePairs(profile, pairs).map(
               createJevCandidate,
             );
@@ -497,22 +498,16 @@ export const registerRouterProvider = (
             } else if (useJev && jev) {
               decision.advisor = 'jev';
               rememberAdvisedTurn(turn, 'jev');
-              const advice = await runJev(
-                {
-                  ...jev,
-                  timeoutMs: Math.min(750, jev.timeoutMs),
-                },
-                {
-                  taskSummary: getBoundedRecentContext(
-                    context,
-                    jev.maxStateChars,
-                  ),
-                  candidates,
-                  profile: profile.jev,
-                  routingDeadline,
-                  signal: options?.signal,
-                },
-              ).catch(() => undefined);
+              const advice = await runJev(jev, {
+                taskSummary: getBoundedRecentContext(
+                  context,
+                  jev.maxStateChars,
+                ),
+                candidates,
+                profile: profile.jev,
+                routingDeadline,
+                signal: options?.signal,
+              }).catch(() => undefined);
               options?.signal?.throwIfAborted();
               // Re-read registry capabilities after the network boundary.
               pairs = available();

package/model-router.example.json

@@ -5,7 +5,7 @@
     "apiKey": "<rendered by chezmoi/1Password in user config only>",
     "endpoint": "https://api.typesafe.ai/v1/systemone",
     "model": "jev-1.13.0",
-    "timeoutMs": 750,
+    "timeoutMs": 1500,
     "confidenceThreshold": 0.65,
     "maxStateChars": 12000,
     "mode": "advisory"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@alexeiled/pi-model-router",
-  "version": "0.6.1",
+  "version": "0.6.2",
   "type": "module",
   "files": [
     "extensions",