release: v3.1.2 — fix critical tool calling regression (#618)

Changes:
- fix: disable proxy_ tool prefix for Claude passthrough (Bash → proxy_Bash)
- docs: document Kiro account ban as upstream AWS issue (#649)
- docs: update CHANGELOG and OpenAPI spec

Fixes #618, closes #649, closes #615

CHANGELOG.md

@@ -4,6 +4,37 @@
 
 ---
 
+## [3.1.2] — 2026-03-26
+
+### 🐛 Bug Fixes
+
+- **Critical: Tool Calling Regression** — Fixed `proxy_Bash` errors by disabling the `proxy_` tool name prefix in the Claude passthrough path. Tools like `Bash`, `Read`, `Write` were being renamed to `proxy_Bash`, `proxy_Read`, etc., causing Claude to reject them (#618)
+- **Kiro Account Ban Documentation** — Documented as upstream AWS anti-fraud false positive, not an OmniRoute issue (#649)
+
+### 🧪 Tests
+
+- **936 tests, 0 failures**
+
+---
+
+## [3.1.1] — 2026-03-26
+
+### ✨ New Features
+
+- **Vision Capability Metadata**: Added `capabilities.vision`, `input_modalities`, and `output_modalities` to `/v1/models` entries for vision-capable models (PR #646)
+- **Gemini 3.1 Models**: Added `gemini-3.1-pro-preview` and `gemini-3.1-flash-lite-preview` to the Antigravity provider (#645)
+
+### 🐛 Bug Fixes
+
+- **Ollama Cloud 401 Error**: Fixed incorrect API base URL — changed from `api.ollama.com` to official `ollama.com/v1/chat/completions` (#643)
+- **Expired Token Retry**: Added bounded retry with exponential backoff (5→10→20 min) for expired OAuth connections instead of permanently skipping them (PR #647)
+
+### 🧪 Tests
+
+- **936 tests, 0 failures**
+
+---
+
 ## [3.1.0] — 2026-03-26
 
 ### ✨ New Features

docs/openapi.yaml

@@ -1,7 +1,7 @@
 openapi: 3.1.0
 info:
   title: OmniRoute API
-  version: 3.1.0
+  version: 3.1.2
   description: |
     OmniRoute is a local-first AI API proxy router. It provides an OpenAI-compatible
     endpoint that routes requests to multiple AI providers with load balancing,

open-sse/config/providerRegistry.ts

@@ -393,6 +393,8 @@ export const REGISTRY: Record<string, RegistryEntry> = {
       { id: "claude-sonnet-4-6", name: "Claude Sonnet 4.6" },
       { id: "claude-sonnet-4-5", name: "Claude Sonnet 4.5" },
       { id: "claude-sonnet-4", name: "Claude Sonnet 4" },
+      { id: "gemini-3.1-pro-preview", name: "Gemini 3.1 Pro Preview" },
+      { id: "gemini-3.1-flash-lite-preview", name: "Gemini 3.1 Flash Lite Preview" },
       { id: "gemini-2.5-pro", name: "Gemini 2.5 Pro" },
       { id: "gemini-2.5-flash", name: "Gemini 2.5 Flash" },
       { id: "gemini-2.0-flash", name: "Gemini 2.0 Flash" },
@@ -1041,8 +1043,8 @@ export const REGISTRY: Record<string, RegistryEntry> = {
     alias: "ollamacloud",
     format: "openai",
     executor: "default",
-    baseUrl: "https://api.ollama.com/v1/chat/completions",
-    modelsUrl: "https://api.ollama.com/v1/models",
+    baseUrl: "https://ollama.com/v1/chat/completions",
+    modelsUrl: "https://ollama.com/api/tags",
     authType: "apikey",
     authHeader: "bearer",
     // Note: rate limits vary by plan (free = "Light usage", Pro = more, Max = 5x Pro).

open-sse/handlers/chatCore.ts

@@ -426,10 +426,12 @@ export async function handleChatCore({
     } else {
       translatedBody = { ...body };
 
-      // Issue #199: Disable tool name prefix when routing Claude-format requests
-      // to non-Claude backends (prefix causes tool name mismatches)
-      const claudeProviders = ["claude", "anthropic"];
-      if (targetFormat === FORMATS.CLAUDE && !claudeProviders.includes(provider?.toLowerCase?.())) {
+      // Issue #199 + #618: Always disable tool name prefix in Claude passthrough.
+      // The proxy_ prefix was designed for OpenAI→Claude translation to avoid
+      // conflicts with Claude OAuth tools, but in the passthrough path the tools
+      // are already in Claude format. Applying the prefix turns "Bash" into
+      // "proxy_Bash", which Claude rejects ("No such tool available: proxy_Bash").
+      if (targetFormat === FORMATS.CLAUDE) {
         translatedBody._disableToolPrefix = true;
       }
 

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "omniroute",
-  "version": "3.1.0",
+  "version": "3.1.2",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "omniroute",
-      "version": "3.1.0",
+      "version": "3.1.2",
       "hasInstallScript": true,
       "license": "MIT",
       "workspaces": [

package.json

@@ -1,6 +1,6 @@
 {
   "name": "omniroute",
-  "version": "3.1.0",
+  "version": "3.1.2",
   "description": "Smart AI Router with auto fallback — route to FREE & cheap models, zero downtime. Works with Cursor, Cline, Claude Desktop, Codex, and any OpenAI-compatible tool.",
   "type": "module",
   "bin": {

src/app/api/v1/models/catalog.ts

@@ -33,6 +33,47 @@ const FALLBACK_ALIAS_TO_PROVIDER = {
   qw: "qwen",
 };
 
+const VISION_MODEL_KEYWORDS = [
+  "gpt-4o",
+  "gpt-4.1",
+  "gpt-4-vision",
+  "gpt-4-turbo",
+  "claude-3",
+  "claude-3.5",
+  "claude-3-5",
+  "claude-4",
+  "claude-opus",
+  "claude-sonnet",
+  "claude-haiku",
+  "gemini",
+  "gemma",
+  "llava",
+  "bakllava",
+  "pixtral",
+  "mistral-pixtral",
+  "qwen-vl",
+  "qvq",
+  "glm-4.6v",
+  "glm-4.5v",
+  "vision",
+  "multimodal",
+];
+
+function isVisionModelId(modelId: string): boolean {
+  const normalized = String(modelId || "").toLowerCase();
+  if (!normalized) return false;
+  return VISION_MODEL_KEYWORDS.some((keyword) => normalized.includes(keyword));
+}
+
+function getVisionCapabilityFields(modelId: string) {
+  if (!isVisionModelId(modelId)) return null;
+  return {
+    capabilities: { vision: true },
+    input_modalities: ["text", "image"],
+    output_modalities: ["text"],
+  };
+}
+
 function buildAliasMaps() {
   const aliasToProviderId: Record<string, string> = {};
   const providerIdToAlias: Record<string, string> = {};
@@ -214,6 +255,8 @@ export async function getUnifiedModelsResponse(
 
       for (const model of providerModels) {
         const aliasId = `${alias}/${model.id}`;
+        const visionFields =
+          getVisionCapabilityFields(aliasId) || getVisionCapabilityFields(model.id);
         // Model-level context length overrides provider default
         const contextLength = model.contextLength || defaultContextLength;
 
@@ -226,13 +269,17 @@ export async function getUnifiedModelsResponse(
           root: model.id,
           parent: null,
           ...(contextLength ? { context_length: contextLength } : {}),
+          ...(visionFields || {}),
         });
 
         // Add provider-id prefix in addition to short alias (ex: kiro/model + kr/model).
         // This improves compatibility for clients that expect full provider names.
         if (canonicalProviderId !== alias) {
+          const providerIdModel = `${canonicalProviderId}/${model.id}`;
+          const providerVisionFields =
+            getVisionCapabilityFields(providerIdModel) || getVisionCapabilityFields(model.id);
           models.push({
-            id: `${canonicalProviderId}/${model.id}`,
+            id: providerIdModel,
             object: "model",
             created: timestamp,
             owned_by: canonicalProviderId,
@@ -240,6 +287,7 @@ export async function getUnifiedModelsResponse(
             root: model.id,
             parent: aliasId,
             ...(contextLength ? { context_length: contextLength } : {}),
+            ...(providerVisionFields || {}),
           });
         }
       }
@@ -383,6 +431,10 @@ export async function getUnifiedModelsResponse(
           if (endpoints.includes("embeddings")) modelType = "embedding";
           else if (endpoints.includes("images")) modelType = "image";
           else if (endpoints.includes("audio")) modelType = "audio";
+          const visionFields =
+            modelType === "chat"
+              ? getVisionCapabilityFields(aliasId) || getVisionCapabilityFields(modelId)
+              : null;
 
           models.push({
             id: aliasId,
@@ -398,12 +450,18 @@ export async function getUnifiedModelsResponse(
             ...(endpoints.length > 1 || !endpoints.includes("chat")
               ? { supported_endpoints: endpoints }
               : {}),
+            ...(visionFields || {}),
           });
 
           // Only add provider-prefixed version if different from alias
           if (canonicalProviderId !== alias && !prefix) {
             const providerPrefixedId = `${canonicalProviderId}/${modelId}`;
             if (models.some((m) => m.id === providerPrefixedId)) continue;
+            const providerVisionFields =
+              modelType === "chat"
+                ? getVisionCapabilityFields(providerPrefixedId) ||
+                  getVisionCapabilityFields(modelId)
+                : null;
             models.push({
               id: providerPrefixedId,
               object: "model",
@@ -414,6 +472,7 @@ export async function getUnifiedModelsResponse(
               parent: aliasId,
               custom: true,
               ...(modelType ? { type: modelType } : {}),
+              ...(providerVisionFields || {}),
             });
           }
         }

src/lib/tokenHealthCheck.ts

@@ -20,6 +20,8 @@ import {
 // ── Constants ────────────────────────────────────────────────────────────────
 const TICK_MS = 60 * 1000; // sweep interval: every 60 seconds
 const DEFAULT_HEALTH_CHECK_INTERVAL_MIN = 60; // default per-connection interval
+const EXPIRED_RETRY_MAX = 3; // max retry attempts for expired connections before giving up
+const EXPIRED_RETRY_BACKOFF_MIN = 5; // backoff between expired retries (minutes)
 const LOG_PREFIX = "[HealthCheck]";
 const TRUE_ENV_VALUES = new Set(["1", "true", "yes", "on"]);
 
@@ -172,8 +174,19 @@ async function checkConnection(conn) {
   if (!conn.isActive) return;
   if (!conn.refreshToken || typeof conn.refreshToken !== "string") return;
 
-  // Skip connections already marked as expired (need re-auth, not retry)
-  if (conn.testStatus === "expired") return;
+  // Retry expired connections with exponential backoff up to EXPIRED_RETRY_MAX times.
+  if (conn.testStatus === "expired") {
+    const retryCount = conn.expiredRetryCount ?? 0;
+    if (retryCount >= EXPIRED_RETRY_MAX) return;
+
+    const lastRetry = conn.expiredRetryAt ? new Date(conn.expiredRetryAt).getTime() : 0;
+    const backoffMs = EXPIRED_RETRY_BACKOFF_MIN * 60 * 1000 * Math.pow(2, retryCount);
+    if (Date.now() - lastRetry < backoffMs) return;
+
+    log(
+      `${LOG_PREFIX} Retrying expired ${conn.provider}/${conn.name || conn.email || conn.id} (attempt ${retryCount + 1}/${EXPIRED_RETRY_MAX})`
+    );
+  }
 
   if (!supportsTokenRefresh(conn.provider)) {
     const now = new Date().toISOString();
@@ -248,7 +261,6 @@ async function checkConnection(conn) {
   }
 
   if (result && result.accessToken) {
-    // Token refreshed successfully — update DB
     const updateData: any = {
       accessToken: result.accessToken,
       lastHealthCheckAt: now,
@@ -258,6 +270,8 @@ async function checkConnection(conn) {
       lastErrorType: null,
       lastErrorSource: null,
       errorCode: null,
+      expiredRetryCount: null,
+      expiredRetryAt: null,
     };
 
     if (result.refreshToken) {
@@ -271,18 +285,22 @@ async function checkConnection(conn) {
     await updateProviderConnection(conn.id, updateData);
     log(`${LOG_PREFIX} ✓ ${conn.provider}/${conn.name || conn.email || conn.id} refreshed`);
   } else {
-    // Refresh failed — record but don't disable the connection
+    const wasExpired = conn.testStatus === "expired";
+    const retryCount = (conn.expiredRetryCount ?? 0) + (wasExpired ? 1 : 0);
+
     await updateProviderConnection(conn.id, {
       lastHealthCheckAt: now,
-      testStatus: "error",
+      testStatus: wasExpired ? "expired" : "error",
       lastError: "Health check: token refresh failed",
       lastErrorAt: now,
       lastErrorType: "token_refresh_failed",
       lastErrorSource: "oauth",
       errorCode: "refresh_failed",
+      ...(wasExpired ? { expiredRetryCount: retryCount, expiredRetryAt: now } : {}),
     });
     logWarn(
-      `${LOG_PREFIX} ✗ ${conn.provider}/${conn.name || conn.email || conn.id} refresh failed`
+      `${LOG_PREFIX} ✗ ${conn.provider}/${conn.name || conn.email || conn.id} refresh failed` +
+        (wasExpired ? ` (expired retry ${retryCount}/${EXPIRED_RETRY_MAX})` : "")
     );
   }
 }