claudish 10.1.0 → 10.1.1

package/dist/index.js

@@ -715,7 +715,7 @@ var init_onepassword_config = __esm(() => {
 });
 
 // src/version.ts
-var VERSION = "10.1.0";
+var VERSION = "10.1.1";
 
 // src/logger.ts
 import { appendFile, existsSync as existsSync2, mkdirSync, readdirSync, unlinkSync, writeFileSync as writeFileSync2 } from "fs";
@@ -18340,7 +18340,6 @@ var init_catalog_route_bindings = __esm(() => {
     "qwen-coding": { routeId: "qwen", routeProfileId: "modelstudio-coding-plan" },
     "qwen-token-plan": { routeId: "qwen", routeProfileId: "qwencloud-token-plan" },
     "qwen-payg": { routeId: "qwen", routeProfileId: "dashscope-direct" },
-    qwen: { routeId: "qwen", routeProfileId: "dashscope-direct" },
     "opencode-zen-go": { routeId: "opencode", routeProfileId: "go-subscription" },
     "opencode-zen": { routeId: "opencode", routeProfileId: "zen" },
     zen: { routeId: "opencode", routeProfileId: "zen" },
@@ -20290,10 +20289,7 @@ class BaseAPIFormat {
     const effort = this.resolveEffortLevel(originalRequest);
     if (!effort)
       return request;
-    if (effort === "none" || effort === "minimal") {
-      if (reasoning?.mandatory) {
-        return this.enableAnthropicEffort(request, effort, reasoning, "mandatory reasoning");
-      }
+    if (this.meansReasoningOff(effort, reasoning)) {
       request.thinking = { type: "disabled" };
       log(`[${this.getName()}] effort ${effort} -> thinking.type: disabled for ${this.modelId}`);
       return request;
@@ -20348,9 +20344,23 @@ class BaseAPIFormat {
     if (Object.keys(request.output_config).length === 0)
       delete request.output_config;
   }
+  meansReasoningOff(effort, reasoning) {
+    if (effort !== "none" && effort !== "minimal")
+      return false;
+    if (reasoning?.mandatory)
+      return false;
+    if (effort === "none")
+      return true;
+    const advertised = (reasoning?.efforts ?? []).filter(isEffortLevel);
+    if (advertised.length === 0)
+      return true;
+    const lowest = advertised.reduce((a, b) => EFFORT_ORDER.indexOf(a) <= EFFORT_ORDER.indexOf(b) ? a : b);
+    return lowest === "none";
+  }
   clampToAdvertisedEffort(requested, reasoning) {
-    if (this.pinnedEffort)
+    if (this.pinnedEffort && this.pinnedEffort !== "none" && this.pinnedEffort !== "minimal") {
       return this.pinnedEffort;
+    }
     const advertised = (reasoning.efforts ?? []).filter(isEffortLevel);
     if (advertised.length === 0) {
       return isEffortLevel(reasoning.defaultEffort) ? reasoning.defaultEffort : undefined;
@@ -20395,6 +20405,9 @@ class BaseAPIFormat {
         return 38912;
       case "max":
         return;
+      case "none":
+      case "minimal":
+        return MIN_THINKING_BUDGET;
       default:
         return 8192;
     }
@@ -23302,7 +23315,7 @@ var init_deepseek_model_dialect = __esm(() => {
     applyNativeReasoning(request, originalRequest) {
       const effort = this.resolveEffortLevel(originalRequest);
       if (effort && this.acceptsReasoningControls()) {
-        if (effort === "none" || effort === "minimal") {
+        if (this.meansReasoningOff(effort, this.lookupReasoningCapability())) {
           request.thinking = { type: "disabled" };
           if (request.reasoning_effort !== undefined)
             delete request.reasoning_effort;
@@ -23359,7 +23372,7 @@ var init_glm_model_dialect = __esm(() => {
       const effort = this.resolveEffortLevel(originalRequest);
       const reasoning = this.lookupReasoningCapability();
       if (effort && this.acceptsThinkingToggle(reasoning)) {
-        if (effort === "none" || effort === "minimal") {
+        if (this.meansReasoningOff(effort, reasoning)) {
           request.thinking = { type: "disabled" };
           if (request.reasoning_effort !== undefined)
             delete request.reasoning_effort;
@@ -23758,7 +23771,7 @@ var init_qwen_model_dialect = __esm(() => {
       const effort = this.resolveEffortLevel(originalRequest);
       if (!effort)
         return request;
-      if (effort === "none" || effort === "minimal") {
+      if (this.meansReasoningOff(effort, this.lookupReasoningCapability())) {
         request.enable_thinking = false;
         log(`[QwenModelDialect] effort ${effort} -> enable_thinking: false for ${this.modelId}`);
       } else {
@@ -25821,8 +25834,8 @@ async function fetchOllamaModels(options = {}) {
     const data = await response.json();
     const models = data.models || [];
     const enriched = await Promise.all(models.map(async (m) => {
-      let capabilities = [];
-      if (enrichCapabilities) {
+      let capabilities = Array.isArray(m.capabilities) ? m.capabilities : [];
+      if (enrichCapabilities && capabilities.length === 0) {
         try {
           const showResponse = await fetch(`${host}/api/show`, {
             method: "POST",
@@ -25836,9 +25849,8 @@ async function fetchOllamaModels(options = {}) {
           }
         } catch {}
       }
-      const nameLower = String(m.name).toLowerCase();
       const supportsTools = capabilities.includes("tools");
-      const isEmbeddingModel = capabilities.includes("embedding") || nameLower.includes("embed");
+      const isEmbeddingModel = capabilities.includes("embedding");
       const sizeInfo = m.details?.parameter_size || "unknown size";
       const toolsIndicator = supportsTools ? "\u2713 tools" : "\u2717 no tools";
       return {
@@ -30287,6 +30299,461 @@ var init_devin = __esm(() => {
   SPEED_SUFFIXES = ["fast", "priority"];
 });
 
+// src/providers/transport/probe-discovery.ts
+function ollamaReported(row) {
+  const caps = Array.isArray(row.capabilities) ? row.capabilities : [];
+  if (caps.includes("completion"))
+    return "chat";
+  if (caps.includes("embedding"))
+    return "not-chat";
+  return;
+}
+function lmStudioReported(row) {
+  if (row.type === "llm" || row.type === "vlm")
+    return "chat";
+  if (row.type === "embeddings" || row.type === "embedding")
+    return "not-chat";
+  return;
+}
+function isSmallName(name) {
+  return SMALL_MODEL_PATTERNS.some((re) => re.test(name));
+}
+function _clearChatCapabilityIndex() {
+  _catalogChatIndex.clear();
+}
+function catalogKey(name) {
+  const lower = name.toLowerCase();
+  return lower.includes("/") ? lower.slice(lower.lastIndexOf("/") + 1) : lower;
+}
+function indexOutputModality(index, keys, modalities) {
+  if (!Array.isArray(modalities) || modalities.length === 0)
+    return;
+  const target = modalities.includes("text") ? index.textOutput : index.nonTextOutput;
+  for (const k of keys)
+    target.add(k);
+}
+function indexInputModality(index, keys, modalities) {
+  if (!Array.isArray(modalities) || modalities.length === 0)
+    return;
+  if (modalities.includes("text"))
+    return;
+  for (const k of keys)
+    index.nonTextInput.add(k);
+}
+function catalogCapabilityIndex(cachePath) {
+  const key = cachePath ?? "";
+  const hit = _catalogChatIndex.get(key);
+  if (hit && hit.expiresAt > Date.now())
+    return hit.index;
+  const index = {
+    chat: new Set,
+    textOutput: new Set,
+    nonTextOutput: new Set,
+    nonTextInput: new Set,
+    videoOutput: new Set,
+    videoOutputKnown: new Set,
+    known: new Set
+  };
+  for (const entry of readAllModelsCache(cachePath)?.entries ?? []) {
+    const keys = [catalogKey(entry.modelId), ...(entry.aliases ?? []).map(catalogKey)];
+    for (const k of keys)
+      index.known.add(k);
+    indexOutputModality(index, keys, entry.outputModalities);
+    indexInputModality(index, keys, entry.inputModalities);
+    if (entry.videoOutput !== undefined) {
+      for (const k of keys)
+        index.videoOutputKnown.add(k);
+    }
+    if (entry.videoOutput === true) {
+      for (const k of keys)
+        index.videoOutput.add(k);
+      continue;
+    }
+    const chatShaped = entry.supportsTools === true || entry.supportsThinking === true;
+    if (chatShaped)
+      for (const k of keys)
+        index.chat.add(k);
+  }
+  _catalogChatIndex.set(key, { index, expiresAt: Date.now() + CATALOG_CHAT_INDEX_TTL_MS });
+  return index;
+}
+function classifyChatCapability(name, cachePath, reported) {
+  if (name.includes("*"))
+    return "not-chat";
+  if (reported)
+    return reported;
+  const index = catalogCapabilityIndex(cachePath);
+  const key = catalogKey(name);
+  if (index.nonTextOutput.has(key))
+    return "not-chat";
+  if (index.nonTextInput.has(key))
+    return "not-chat";
+  if (index.textOutput.has(key))
+    return "chat";
+  if (index.videoOutput.has(key))
+    return "not-chat";
+  if (index.chat.has(key))
+    return "chat";
+  return "unknown";
+}
+function isChatCapable(name) {
+  return classifyChatCapability(name) === "chat";
+}
+function isReportedChatCapable(name, reported) {
+  return classifyChatCapability(name, undefined, reported) === "chat";
+}
+function unavailableForMissingCapability(names, reportedFor, cachePath) {
+  const catalogSilent = [];
+  const providerSilent = [];
+  for (const name of names) {
+    const reported = reportedFor?.(name);
+    if (classifyChatCapability(name, cachePath, reported) !== "unknown")
+      continue;
+    const index = catalogCapabilityIndex(cachePath);
+    if (index.known.has(catalogKey(name)))
+      catalogSilent.push(name);
+    else
+      providerSilent.push(name);
+  }
+  return { catalogSilent, providerSilent };
+}
+function describeNoCandidates(ids, reportedFor, cachePath) {
+  const { catalogSilent, providerSilent } = unavailableForMissingCapability(ids, reportedFor, cachePath);
+  const undescribed = catalogSilent.length + providerSilent.length;
+  if (undescribed === 0) {
+    return `all ${ids.length} listed models are described as non-chat (image, embedding, audio)`;
+  }
+  const sample = [...catalogSilent, ...providerSilent].slice(0, 3).join(", ");
+  const more = undescribed > 3 ? `, +${undescribed - 3} more` : "";
+  const whose = providerSilent.length === 0 ? "the models catalog publishes them without modalities" : catalogSilent.length === 0 ? "this endpoint publishes no capability field" : "neither the models catalog nor this endpoint describes them";
+  return `no capability data for ${undescribed} of ${ids.length} listed models \u2014 ${whose} (${sample}${more})`;
+}
+function isStandardName(name) {
+  if (!name.includes("/"))
+    return true;
+  return STANDARD_VENDOR_PREFIXES.some((p) => name.toLowerCase().startsWith(p));
+}
+function rankProbeCandidates(names) {
+  return names.filter((name) => !name.includes("*")).sort((a, b) => {
+    const aStd = isStandardName(a);
+    const bStd = isStandardName(b);
+    if (aStd !== bStd)
+      return aStd ? -1 : 1;
+    const aSmall = isSmallName(a);
+    const bSmall = isSmallName(b);
+    if (aSmall !== bSmall)
+      return aSmall ? -1 : 1;
+    return a.localeCompare(b);
+  });
+}
+function cacheGet(key, exclude = new Set) {
+  const hit = _cache.get(key);
+  if (!hit)
+    return;
+  if (Date.now() > hit.expiresAt) {
+    _cache.delete(key);
+    return;
+  }
+  if (hit.ranked.length === 0) {
+    return { model: null, reason: hit.reason };
+  }
+  const pick = hit.ranked.find((m) => !exclude.has(m));
+  if (!pick) {
+    return {
+      model: null,
+      reason: `all ${hit.ranked.length} candidate model(s) already tried`
+    };
+  }
+  return { model: pick };
+}
+function cacheSetFailure(key, reason) {
+  _cache.set(key, { ranked: [], reason, expiresAt: Date.now() + CACHE_TTL_MS });
+}
+function cacheSetRanked(key, ranked) {
+  _cache.set(key, { ranked, expiresAt: Date.now() + CACHE_TTL_MS });
+}
+async function discoverViaOpenAIModels(endpoint, headers, cacheKey) {
+  const cached = cacheGet(cacheKey.key, cacheKey.exclude);
+  if (cached !== undefined)
+    return cached;
+  let response;
+  try {
+    response = await fetch(endpoint, {
+      method: "GET",
+      headers,
+      signal: AbortSignal.timeout(FETCH_TIMEOUT_MS)
+    });
+  } catch (e) {
+    const reason = classifyFetchError(e, endpoint);
+    log(`[probe-discovery${cacheKey.displayName ? `:${cacheKey.displayName}` : ""}] fetch failed: ${reason}`);
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  if (!response.ok) {
+    const who = cacheKey.displayName ?? "this provider";
+    const authFailure = response.status === 401 || response.status === 403 ? cacheKey.hasApiKey === false ? `the server requires an API key and none is configured for ${who}` : `the server rejected the configured API key for ${who}` : "";
+    const reason = authFailure ? `HTTP ${response.status} from ${endpoint} \u2014 ${authFailure}` : `HTTP ${response.status} from ${endpoint}`;
+    log(`[probe-discovery${cacheKey.displayName ? `:${cacheKey.displayName}` : ""}] ${reason}`);
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  let body;
+  try {
+    body = await response.json();
+  } catch {
+    const reason = "invalid /v1/models response (not JSON)";
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  const ids = extractModelIds(body);
+  if (ids.length === 0) {
+    const url = tryParseUrl(endpoint);
+    const host = url?.host ?? endpoint;
+    const reason = `${host} reachable but no models loaded \u2014 load a model in the server UI`;
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  const ranked = rankProbeCandidates(ids.filter(isChatCapable));
+  if (ranked.length === 0) {
+    const reason = describeNoCandidates(ids);
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  cacheSetRanked(cacheKey.key, ranked);
+  const pick = ranked.find((m) => !cacheKey.exclude?.has(m));
+  if (!pick) {
+    return {
+      model: null,
+      reason: `all ${ranked.length} candidate model(s) already tried`
+    };
+  }
+  return { model: pick };
+}
+function classifyFetchError(e, endpoint) {
+  const name = e?.name ?? "";
+  const code = e?.cause?.code ?? "";
+  const msg = e instanceof Error ? e.message : String(e);
+  const url = tryParseUrl(endpoint);
+  const host = url?.host ?? endpoint;
+  const isLocal = !!url && /^(localhost|127\.0\.0\.1|0\.0\.0\.0|::1)$/i.test(url.hostname);
+  if (name === "TimeoutError" || name === "AbortError" || /timeout/i.test(msg)) {
+    return `${host} unresponsive (>${FETCH_TIMEOUT_MS / 1000}s) \u2014 check if the server is overloaded`;
+  }
+  if (code === "ENOTFOUND" || code === "EAI_AGAIN") {
+    return `cannot resolve host ${url?.hostname ?? endpoint} \u2014 check the URL`;
+  }
+  const isConnRefused = code === "ECONNREFUSED" || code === "ECONNRESET" || /unable to connect|connection refused|fetch failed/i.test(msg);
+  if (isConnRefused) {
+    if (isLocal) {
+      return `${host} not reachable \u2014 is the server running? Press u to change URL.`;
+    }
+    return `${host} not reachable \u2014 check the URL or network. Press u to change.`;
+  }
+  return `${host}: ${msg}`;
+}
+function tryParseUrl(s) {
+  try {
+    return new URL(s);
+  } catch {
+    return null;
+  }
+}
+function extractModelIds(body) {
+  if (!body || typeof body !== "object")
+    return [];
+  const data = body;
+  if (Array.isArray(data.data)) {
+    return data.data.map((m) => m && typeof m === "object" ? m.id : null).filter((id) => typeof id === "string" && id.length > 0);
+  }
+  if (Array.isArray(data.models)) {
+    return data.models.map((m) => m && typeof m === "object" ? m.id ?? m.model_name : null).filter((id) => typeof id === "string" && id.length > 0);
+  }
+  return [];
+}
+function orderByCost(models) {
+  const sized = models.filter((m) => typeof m.size === "number");
+  const unsized = models.filter((m) => typeof m.size !== "number");
+  const bySize = [...sized].sort((a, b) => (a.size ?? Number.POSITIVE_INFINITY) - (b.size ?? Number.POSITIVE_INFINITY)).map((m) => m.name);
+  return [...bySize, ...rankProbeCandidates(unsized.map((m) => m.name))];
+}
+async function discoverViaOllama(baseUrl, cacheKey) {
+  const cached = cacheGet(cacheKey.key, cacheKey.exclude);
+  if (cached !== undefined)
+    return cached;
+  const [psResult, tagsResult] = await Promise.allSettled([
+    fetchOllamaModels2(`${baseUrl}/api/ps`),
+    fetchOllamaModels2(`${baseUrl}/api/tags`)
+  ]);
+  const loadedRaw = psResult.status === "fulfilled" ? psResult.value : [];
+  const tagsRaw = tagsResult.status === "fulfilled" ? tagsResult.value : [];
+  const connectionError = psResult.status === "rejected" ? classifyFetchError(psResult.reason, `${baseUrl}/api/ps`) : tagsResult.status === "rejected" ? classifyFetchError(tagsResult.reason, `${baseUrl}/api/tags`) : undefined;
+  const loaded = loadedRaw.filter((m) => isReportedChatCapable(m.name, ollamaReported(m)));
+  const loadedNames = new Set(loaded.map((m) => m.name));
+  const rest = tagsRaw.filter((m) => isReportedChatCapable(m.name, ollamaReported(m)) && !loadedNames.has(m.name));
+  if (loaded.length === 0 && rest.length === 0) {
+    const listed = [...loadedRaw, ...tagsRaw];
+    const reportedFor = (name) => {
+      const row = listed.find((m) => m.name === name);
+      return row ? ollamaReported(row) : undefined;
+    };
+    const reason = connectionError ?? (listed.length === 0 ? `no models on ${baseUrl} (pull one: ollama pull llama3.2)` : `${describeNoCandidates(listed.map((m) => m.name), reportedFor)} on ${baseUrl}`);
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  const ranked = [...orderByCost(loaded), ...orderByCost(rest)];
+  if (ranked.length === 0) {
+    const reason = "no chat-capable model on Ollama endpoint";
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  cacheSetRanked(cacheKey.key, ranked);
+  const pick = ranked.find((m) => !cacheKey.exclude?.has(m));
+  if (!pick) {
+    return {
+      model: null,
+      reason: `all ${ranked.length} candidate model(s) already tried`
+    };
+  }
+  return { model: pick };
+}
+async function discoverViaLMStudio(baseUrl, headers, cacheKey) {
+  const cached = cacheGet(cacheKey.key, cacheKey.exclude);
+  if (cached !== undefined)
+    return cached;
+  let response;
+  try {
+    response = await fetch(`${baseUrl}/api/v0/models`, {
+      method: "GET",
+      headers,
+      signal: AbortSignal.timeout(FETCH_TIMEOUT_MS)
+    });
+  } catch (e) {
+    return discoverViaOpenAIModels(`${baseUrl}/v1/models`, headers, cacheKey);
+  }
+  if (!response.ok) {
+    return discoverViaOpenAIModels(`${baseUrl}/v1/models`, headers, cacheKey);
+  }
+  let body;
+  try {
+    body = await response.json();
+  } catch {
+    const reason = "invalid /api/v0/models response (not JSON)";
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  const models = extractLMStudioModels(body);
+  if (models.length === 0) {
+    const url = tryParseUrl(baseUrl);
+    const host = url?.host ?? baseUrl;
+    const reason = `${host} reachable but no models present \u2014 download one in the LM Studio UI`;
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  const chatModels = models.filter((m) => isReportedChatCapable(m.id, lmStudioReported(m)));
+  const loaded = chatModels.filter((m) => m.state === "loaded");
+  const notLoaded = chatModels.filter((m) => m.state !== "loaded");
+  const ranked = [
+    ...rankProbeCandidates(loaded.map((m) => m.id)),
+    ...rankProbeCandidates(notLoaded.map((m) => m.id))
+  ];
+  if (ranked.length === 0) {
+    const url = tryParseUrl(baseUrl);
+    const host = url?.host ?? baseUrl;
+    const reason = `${host} has ${models.length} model(s) but none are chat-capable`;
+    cacheSetFailure(cacheKey.key, reason);
+    return { model: null, reason };
+  }
+  cacheSetRanked(cacheKey.key, ranked);
+  const pick = ranked.find((m) => !cacheKey.exclude?.has(m));
+  if (!pick) {
+    return {
+      model: null,
+      reason: `all ${ranked.length} candidate model(s) already tried`
+    };
+  }
+  return { model: pick };
+}
+function extractLMStudioModels(body) {
+  if (!body || typeof body !== "object")
+    return [];
+  const data = body.data;
+  if (!Array.isArray(data))
+    return [];
+  const out = [];
+  for (const m of data) {
+    if (!m || typeof m !== "object")
+      continue;
+    const r = m;
+    if (typeof r.id !== "string" || !r.id)
+      continue;
+    out.push({
+      id: r.id,
+      state: typeof r.state === "string" ? r.state : undefined,
+      type: typeof r.type === "string" ? r.type : undefined
+    });
+  }
+  return out;
+}
+async function fetchOllamaModels2(url) {
+  const response = await fetch(url, {
+    method: "GET",
+    signal: AbortSignal.timeout(FETCH_TIMEOUT_MS)
+  });
+  if (!response.ok)
+    return [];
+  const body = await response.json().catch(() => null);
+  if (!body?.models)
+    return [];
+  return body.models.map((m) => ({
+    name: typeof m.name === "string" ? m.name : "",
+    size: typeof m.size === "number" ? m.size : undefined,
+    capabilities: Array.isArray(m.capabilities) ? m.capabilities.filter((c) => typeof c === "string") : undefined
+  })).filter((m) => m.name.length > 0);
+}
+function invalidateProbeDiscovery(providerSlug) {
+  for (const key of _cache.keys()) {
+    if (key.startsWith(`${providerSlug}:`)) {
+      _cache.delete(key);
+    }
+  }
+}
+var _cache, CACHE_TTL_MS, FETCH_TIMEOUT_MS = 5000, SMALL_MODEL_PATTERNS, CATALOG_CHAT_INDEX_TTL_MS = 60000, _catalogChatIndex, STANDARD_VENDOR_PREFIXES;
+var init_probe_discovery = __esm(() => {
+  init_logger();
+  init_all_models_cache();
+  _cache = new Map;
+  CACHE_TTL_MS = 5 * 60 * 1000;
+  SMALL_MODEL_PATTERNS = [
+    /\bmini\b/i,
+    /\bnano\b/i,
+    /\bflash\b/i,
+    /\blite\b/i,
+    /\bhaiku\b/i,
+    /\bsmall\b/i,
+    /\btiny\b/i,
+    /\b[12345]b\b/i,
+    /\b[78]b\b/i
+  ];
+  _catalogChatIndex = new Map;
+  STANDARD_VENDOR_PREFIXES = [
+    "openai/",
+    "anthropic/",
+    "google/",
+    "gemini/",
+    "meta/",
+    "meta-llama/",
+    "mistralai/",
+    "mistral/",
+    "x-ai/",
+    "deepseek/",
+    "qwen/",
+    "moonshot/",
+    "moonshotai/",
+    "zhipuai/",
+    "z-ai/"
+  ];
+});
+
 // src/providers/model-discovery-builtins.ts
 function oneLine(text, max = 200) {
   const flat = text.replace(/\s+/g, " ").trim();
@@ -30317,7 +30784,7 @@ async function fetchDevinModelsCatalog() {
       endpoint,
       models: served.map((model) => {
         const { wireId, ...rest } = devinModelsCatalogEntry(model);
-        return { id: wireId, ...rest };
+        return { id: wireId, ...rest, reported: "chat" };
       })
     };
   } catch (err) {
@@ -30356,7 +30823,12 @@ async function fetchAntigravityModelsCatalog() {
       endpoint,
       models: selectable.map((id) => {
         const m = meta[id];
-        return m?.contextWindow ? { id, contextWindow: m.contextWindow, ignoreCatalogReleaseDate: true } : { id, ignoreCatalogReleaseDate: true };
+        return m?.contextWindow ? {
+          id,
+          contextWindow: m.contextWindow,
+          ignoreCatalogReleaseDate: true,
+          reported: "chat"
+        } : { id, ignoreCatalogReleaseDate: true, reported: "chat" };
       })
     };
   } catch (err) {
@@ -30368,19 +30840,18 @@ async function fetchAntigravityModelsCatalog() {
 }
 async function fetchOllamaModelsCatalog() {
   await Promise.resolve();
+  await Promise.resolve().then(() => init_probe_discovery());
   const endpoint = `${ollamaBaseUrl()}/api/tags`;
   try {
-    const installed = await fetchOllamaModels({
-      enrichCapabilities: false,
-      throwOnError: true
-    });
+    const installed = await fetchOllamaModels({ throwOnError: true });
     return {
       kind: "models",
       endpoint,
       models: installed.map((model) => ({
         id: model.name,
         displayName: model.name,
-        supportsTools: model.supportsTools
+        supportsTools: model.supportsTools,
+        reported: ollamaReported(model)
       }))
     };
   } catch (err) {
@@ -30558,11 +31029,11 @@ async function discoverViaFetcher(providerName, format) {
   }
   _failures.delete(providerName);
   log(`[model-discovery:${providerName}] discovered ${models.length} models`);
-  _cache.set(providerName, { models, expiresAt: Date.now() + CACHE_TTL_MS });
+  _cache2.set(providerName, { models, expiresAt: Date.now() + CACHE_TTL_MS2 });
   return { kind: "served", models };
 }
 async function discoverProviderModelsCatalog(providerName) {
-  const cached = _cache.get(providerName);
+  const cached = _cache2.get(providerName);
   if (cached && cached.expiresAt > Date.now())
     return { kind: "served", models: cached.models };
   const def = getProviderByName(providerName);
@@ -30613,7 +31084,7 @@ async function discoverProviderModelsCatalog(providerName) {
     response = await fetch(endpoint, {
       method: "GET",
       headers,
-      signal: AbortSignal.timeout(FETCH_TIMEOUT_MS)
+      signal: AbortSignal.timeout(FETCH_TIMEOUT_MS2)
     });
   } catch (e) {
     return recordFailure({
@@ -30656,7 +31127,7 @@ async function discoverProviderModelsCatalog(providerName) {
   const models = parsed.models;
   _failures.delete(providerName);
   log(`[model-discovery:${providerName}] discovered ${models.length} models: ` + models.map((m) => `${m.id}(${m.contextWindow ?? "?"})`).join(", "));
-  _cache.set(providerName, { models, expiresAt: Date.now() + CACHE_TTL_MS });
+  _cache2.set(providerName, { models, expiresAt: Date.now() + CACHE_TTL_MS2 });
   return { kind: "served", models };
 }
 async function discoverProviderModels(providerName) {
@@ -30674,438 +31145,18 @@ function rankDiscoveredModels(models) {
     return diff !== 0 ? diff : compareByReleaseDateDesc(a, b);
   });
 }
-var _fetchers, CACHE_TTL_MS, FETCH_TIMEOUT_MS = 5000, _cache, _failures, MIN_CREATED_SECONDS = 946684800, MAX_CREATED_SECONDS = 4102444800, CONTINUATION_FIELDS;
+var _fetchers, CACHE_TTL_MS2, FETCH_TIMEOUT_MS2 = 5000, _cache2, _failures, MIN_CREATED_SECONDS = 946684800, MAX_CREATED_SECONDS = 4102444800, CONTINUATION_FIELDS;
 var init_model_discovery = __esm(() => {
   init_authority();
   init_logger();
   init_provider_definitions();
   _fetchers = new Map;
-  CACHE_TTL_MS = 5 * 60 * 1000;
-  _cache = new Map;
+  CACHE_TTL_MS2 = 5 * 60 * 1000;
+  _cache2 = new Map;
   _failures = new Map;
   CONTINUATION_FIELDS = ["has_more", "next", "next_page", "next_page_token"];
 });
 
-// src/providers/transport/probe-discovery.ts
-function isSmallName(name) {
-  return SMALL_MODEL_PATTERNS.some((re) => re.test(name));
-}
-function catalogKey(name) {
-  const lower = name.toLowerCase();
-  return lower.includes("/") ? lower.slice(lower.lastIndexOf("/") + 1) : lower;
-}
-function indexOutputModality(index, keys, modalities) {
-  if (!Array.isArray(modalities) || modalities.length === 0)
-    return;
-  const target = modalities.includes("text") ? index.textOutput : index.nonTextOutput;
-  for (const k of keys)
-    target.add(k);
-}
-function catalogCapabilityIndex(cachePath) {
-  const key = cachePath ?? "";
-  const hit = _catalogChatIndex.get(key);
-  if (hit && hit.expiresAt > Date.now())
-    return hit.index;
-  const index = {
-    chat: new Set,
-    textOutput: new Set,
-    nonTextOutput: new Set,
-    videoOutput: new Set,
-    videoOutputKnown: new Set
-  };
-  for (const entry of readAllModelsCache(cachePath)?.entries ?? []) {
-    const keys = [catalogKey(entry.modelId), ...(entry.aliases ?? []).map(catalogKey)];
-    indexOutputModality(index, keys, entry.outputModalities);
-    if (entry.videoOutput !== undefined) {
-      for (const k of keys)
-        index.videoOutputKnown.add(k);
-    }
-    if (entry.videoOutput === true) {
-      for (const k of keys)
-        index.videoOutput.add(k);
-      continue;
-    }
-    const chatShaped = entry.supportsTools === true || entry.supportsThinking === true || entry.supportsVision === true;
-    if (chatShaped)
-      for (const k of keys)
-        index.chat.add(k);
-  }
-  _catalogChatIndex.set(key, { index, expiresAt: Date.now() + CATALOG_CHAT_INDEX_TTL_MS });
-  return index;
-}
-function classifyChatCapability(name, cachePath) {
-  if (name.includes("*"))
-    return "not-chat";
-  const index = catalogCapabilityIndex(cachePath);
-  const key = catalogKey(name);
-  if (index.nonTextOutput.has(key))
-    return "not-chat";
-  if (index.textOutput.has(key))
-    return "chat";
-  if (index.videoOutput.has(key))
-    return "not-chat";
-  if (NON_CHAT_PATTERNS.some((re) => re.test(name)))
-    return "not-chat";
-  if (!index.videoOutputKnown.has(key) && VIDEO_OUTPUT_NAME_PATTERNS.some((re) => re.test(name))) {
-    return "not-chat";
-  }
-  if (index.chat.has(key))
-    return "chat";
-  return "unknown";
-}
-function isChatCapable(name) {
-  return classifyChatCapability(name) !== "not-chat";
-}
-function isStandardName(name) {
-  if (!name.includes("/"))
-    return true;
-  return STANDARD_VENDOR_PREFIXES.some((p) => name.toLowerCase().startsWith(p));
-}
-function rankProbeCandidates(names) {
-  return names.filter(isChatCapable).sort((a, b) => {
-    const aStd = isStandardName(a);
-    const bStd = isStandardName(b);
-    if (aStd !== bStd)
-      return aStd ? -1 : 1;
-    const aSmall = isSmallName(a);
-    const bSmall = isSmallName(b);
-    if (aSmall !== bSmall)
-      return aSmall ? -1 : 1;
-    return a.localeCompare(b);
-  });
-}
-function cacheGet(key, exclude = new Set) {
-  const hit = _cache2.get(key);
-  if (!hit)
-    return;
-  if (Date.now() > hit.expiresAt) {
-    _cache2.delete(key);
-    return;
-  }
-  if (hit.ranked.length === 0) {
-    return { model: null, reason: hit.reason };
-  }
-  const pick = hit.ranked.find((m) => !exclude.has(m));
-  if (!pick) {
-    return {
-      model: null,
-      reason: `all ${hit.ranked.length} candidate model(s) already tried`
-    };
-  }
-  return { model: pick };
-}
-function cacheSetFailure(key, reason) {
-  _cache2.set(key, { ranked: [], reason, expiresAt: Date.now() + CACHE_TTL_MS2 });
-}
-function cacheSetRanked(key, ranked) {
-  _cache2.set(key, { ranked, expiresAt: Date.now() + CACHE_TTL_MS2 });
-}
-async function discoverViaOpenAIModels(endpoint, headers, cacheKey) {
-  const cached = cacheGet(cacheKey.key, cacheKey.exclude);
-  if (cached !== undefined)
-    return cached;
-  let response;
-  try {
-    response = await fetch(endpoint, {
-      method: "GET",
-      headers,
-      signal: AbortSignal.timeout(FETCH_TIMEOUT_MS2)
-    });
-  } catch (e) {
-    const reason = classifyFetchError(e, endpoint);
-    log(`[probe-discovery${cacheKey.displayName ? `:${cacheKey.displayName}` : ""}] fetch failed: ${reason}`);
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  if (!response.ok) {
-    const who = cacheKey.displayName ?? "this provider";
-    const authFailure = response.status === 401 || response.status === 403 ? cacheKey.hasApiKey === false ? `the server requires an API key and none is configured for ${who}` : `the server rejected the configured API key for ${who}` : "";
-    const reason = authFailure ? `HTTP ${response.status} from ${endpoint} \u2014 ${authFailure}` : `HTTP ${response.status} from ${endpoint}`;
-    log(`[probe-discovery${cacheKey.displayName ? `:${cacheKey.displayName}` : ""}] ${reason}`);
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  let body;
-  try {
-    body = await response.json();
-  } catch {
-    const reason = "invalid /v1/models response (not JSON)";
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  const ids = extractModelIds(body);
-  if (ids.length === 0) {
-    const url = tryParseUrl(endpoint);
-    const host = url?.host ?? endpoint;
-    const reason = `${host} reachable but no models loaded \u2014 load a model in the server UI`;
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  const ranked = rankProbeCandidates(ids);
-  if (ranked.length === 0) {
-    const reason = `no chat-capable model among ${ids.length} listed`;
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  cacheSetRanked(cacheKey.key, ranked);
-  const pick = ranked.find((m) => !cacheKey.exclude?.has(m));
-  if (!pick) {
-    return {
-      model: null,
-      reason: `all ${ranked.length} candidate model(s) already tried`
-    };
-  }
-  return { model: pick };
-}
-function classifyFetchError(e, endpoint) {
-  const name = e?.name ?? "";
-  const code = e?.cause?.code ?? "";
-  const msg = e instanceof Error ? e.message : String(e);
-  const url = tryParseUrl(endpoint);
-  const host = url?.host ?? endpoint;
-  const isLocal = !!url && /^(localhost|127\.0\.0\.1|0\.0\.0\.0|::1)$/i.test(url.hostname);
-  if (name === "TimeoutError" || name === "AbortError" || /timeout/i.test(msg)) {
-    return `${host} unresponsive (>${FETCH_TIMEOUT_MS2 / 1000}s) \u2014 check if the server is overloaded`;
-  }
-  if (code === "ENOTFOUND" || code === "EAI_AGAIN") {
-    return `cannot resolve host ${url?.hostname ?? endpoint} \u2014 check the URL`;
-  }
-  const isConnRefused = code === "ECONNREFUSED" || code === "ECONNRESET" || /unable to connect|connection refused|fetch failed/i.test(msg);
-  if (isConnRefused) {
-    if (isLocal) {
-      return `${host} not reachable \u2014 is the server running? Press u to change URL.`;
-    }
-    return `${host} not reachable \u2014 check the URL or network. Press u to change.`;
-  }
-  return `${host}: ${msg}`;
-}
-function tryParseUrl(s) {
-  try {
-    return new URL(s);
-  } catch {
-    return null;
-  }
-}
-function extractModelIds(body) {
-  if (!body || typeof body !== "object")
-    return [];
-  const data = body;
-  if (Array.isArray(data.data)) {
-    return data.data.map((m) => m && typeof m === "object" ? m.id : null).filter((id) => typeof id === "string" && id.length > 0);
-  }
-  if (Array.isArray(data.models)) {
-    return data.models.map((m) => m && typeof m === "object" ? m.id ?? m.model_name : null).filter((id) => typeof id === "string" && id.length > 0);
-  }
-  return [];
-}
-function orderByCost(models) {
-  const sized = models.filter((m) => typeof m.size === "number");
-  const unsized = models.filter((m) => typeof m.size !== "number");
-  const bySize = [...sized].sort((a, b) => (a.size ?? Number.POSITIVE_INFINITY) - (b.size ?? Number.POSITIVE_INFINITY)).map((m) => m.name);
-  return [...bySize, ...rankProbeCandidates(unsized.map((m) => m.name))];
-}
-async function discoverViaOllama(baseUrl, cacheKey) {
-  const cached = cacheGet(cacheKey.key, cacheKey.exclude);
-  if (cached !== undefined)
-    return cached;
-  const [psResult, tagsResult] = await Promise.allSettled([
-    fetchOllamaModels2(`${baseUrl}/api/ps`),
-    fetchOllamaModels2(`${baseUrl}/api/tags`)
-  ]);
-  const loadedRaw = psResult.status === "fulfilled" ? psResult.value : [];
-  const tagsRaw = tagsResult.status === "fulfilled" ? tagsResult.value : [];
-  const connectionError = psResult.status === "rejected" ? classifyFetchError(psResult.reason, `${baseUrl}/api/ps`) : tagsResult.status === "rejected" ? classifyFetchError(tagsResult.reason, `${baseUrl}/api/tags`) : undefined;
-  const loaded = loadedRaw.filter((m) => isChatCapable(m.name));
-  const loadedNames = new Set(loaded.map((m) => m.name));
-  const rest = tagsRaw.filter((m) => isChatCapable(m.name) && !loadedNames.has(m.name));
-  if (loaded.length === 0 && rest.length === 0) {
-    const reason = connectionError ?? (loadedRaw.length === 0 && tagsRaw.length === 0 ? `no models on ${baseUrl} (pull one: ollama pull llama3.2)` : `only embedding/non-chat models on ${baseUrl}`);
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  const ranked = [...orderByCost(loaded), ...orderByCost(rest)];
-  if (ranked.length === 0) {
-    const reason = "no chat-capable model on Ollama endpoint";
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  cacheSetRanked(cacheKey.key, ranked);
-  const pick = ranked.find((m) => !cacheKey.exclude?.has(m));
-  if (!pick) {
-    return {
-      model: null,
-      reason: `all ${ranked.length} candidate model(s) already tried`
-    };
-  }
-  return { model: pick };
-}
-async function discoverViaLMStudio(baseUrl, headers, cacheKey) {
-  const cached = cacheGet(cacheKey.key, cacheKey.exclude);
-  if (cached !== undefined)
-    return cached;
-  let response;
-  try {
-    response = await fetch(`${baseUrl}/api/v0/models`, {
-      method: "GET",
-      headers,
-      signal: AbortSignal.timeout(FETCH_TIMEOUT_MS2)
-    });
-  } catch (e) {
-    return discoverViaOpenAIModels(`${baseUrl}/v1/models`, headers, cacheKey);
-  }
-  if (!response.ok) {
-    return discoverViaOpenAIModels(`${baseUrl}/v1/models`, headers, cacheKey);
-  }
-  let body;
-  try {
-    body = await response.json();
-  } catch {
-    const reason = "invalid /api/v0/models response (not JSON)";
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  const models = extractLMStudioModels(body);
-  if (models.length === 0) {
-    const url = tryParseUrl(baseUrl);
-    const host = url?.host ?? baseUrl;
-    const reason = `${host} reachable but no models present \u2014 download one in the LM Studio UI`;
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  const chatModels = models.filter((m) => isChatCapable(m.id) && m.type !== "embeddings" && m.type !== "embedding");
-  const loaded = chatModels.filter((m) => m.state === "loaded");
-  const notLoaded = chatModels.filter((m) => m.state !== "loaded");
-  const ranked = [
-    ...rankProbeCandidates(loaded.map((m) => m.id)),
-    ...rankProbeCandidates(notLoaded.map((m) => m.id))
-  ];
-  if (ranked.length === 0) {
-    const url = tryParseUrl(baseUrl);
-    const host = url?.host ?? baseUrl;
-    const reason = `${host} has ${models.length} model(s) but none are chat-capable`;
-    cacheSetFailure(cacheKey.key, reason);
-    return { model: null, reason };
-  }
-  cacheSetRanked(cacheKey.key, ranked);
-  const pick = ranked.find((m) => !cacheKey.exclude?.has(m));
-  if (!pick) {
-    return {
-      model: null,
-      reason: `all ${ranked.length} candidate model(s) already tried`
-    };
-  }
-  return { model: pick };
-}
-function extractLMStudioModels(body) {
-  if (!body || typeof body !== "object")
-    return [];
-  const data = body.data;
-  if (!Array.isArray(data))
-    return [];
-  const out = [];
-  for (const m of data) {
-    if (!m || typeof m !== "object")
-      continue;
-    const r = m;
-    if (typeof r.id !== "string" || !r.id)
-      continue;
-    out.push({
-      id: r.id,
-      state: typeof r.state === "string" ? r.state : undefined,
-      type: typeof r.type === "string" ? r.type : undefined
-    });
-  }
-  return out;
-}
-async function fetchOllamaModels2(url) {
-  const response = await fetch(url, {
-    method: "GET",
-    signal: AbortSignal.timeout(FETCH_TIMEOUT_MS2)
-  });
-  if (!response.ok)
-    return [];
-  const body = await response.json().catch(() => null);
-  if (!body?.models)
-    return [];
-  return body.models.map((m) => ({
-    name: typeof m.name === "string" ? m.name : "",
-    size: typeof m.size === "number" ? m.size : undefined
-  })).filter((m) => m.name.length > 0);
-}
-function invalidateProbeDiscovery(providerSlug) {
-  for (const key of _cache2.keys()) {
-    if (key.startsWith(`${providerSlug}:`)) {
-      _cache2.delete(key);
-    }
-  }
-}
-var _cache2, CACHE_TTL_MS2, FETCH_TIMEOUT_MS2 = 5000, SMALL_MODEL_PATTERNS, NON_CHAT_PATTERNS, VIDEO_OUTPUT_NAME_PATTERNS, CATALOG_CHAT_INDEX_TTL_MS = 60000, _catalogChatIndex, STANDARD_VENDOR_PREFIXES;
-var init_probe_discovery = __esm(() => {
-  init_logger();
-  init_all_models_cache();
-  _cache2 = new Map;
-  CACHE_TTL_MS2 = 5 * 60 * 1000;
-  SMALL_MODEL_PATTERNS = [
-    /\bmini\b/i,
-    /\bnano\b/i,
-    /\bflash\b/i,
-    /\blite\b/i,
-    /\bhaiku\b/i,
-    /\bsmall\b/i,
-    /\btiny\b/i,
-    /\b[12345]b\b/i,
-    /\b[78]b\b/i
-  ];
-  NON_CHAT_PATTERNS = [
-    /\bimage\b/i,
-    /\bembed/i,
-    /\bminilm\b/i,
-    /\bnomic-embed/i,
-    /\bbge-/i,
-    /\bmxbai-embed/i,
-    /\btts\b/i,
-    /\bwhisper\b/i,
-    /\baudio\b/i,
-    /\bvoxtral\b/i,
-    /\bdall-?e\b/i,
-    /\bmoderation\b/i,
-    /\brerank/i,
-    /\bspeech\b/i,
-    /\btranscribe\b/i,
-    /\btranscription\b/i,
-    /\bvoice\b/i,
-    /\brealtime\b/i,
-    /\blive\b/i,
-    /\btranslate\b/i,
-    /\btranslation\b/i,
-    /-(image|tts|audio|embedding|vision-only|transcribe|voice|speech|realtime|live|translate)(-|$)/i
-  ];
-  VIDEO_OUTPUT_NAME_PATTERNS = [
-    /\bvideo\b/i,
-    /(^|[-_.])(t2v|i2v|r2v|v2v)([-_.]|$)/i,
-    /\bveo\b/i,
-    /\bsora\b/i
-  ];
-  _catalogChatIndex = new Map;
-  STANDARD_VENDOR_PREFIXES = [
-    "openai/",
-    "anthropic/",
-    "google/",
-    "gemini/",
-    "meta/",
-    "meta-llama/",
-    "mistralai/",
-    "mistral/",
-    "x-ai/",
-    "deepseek/",
-    "qwen/",
-    "moonshot/",
-    "moonshotai/",
-    "zhipuai/",
-    "z-ai/"
-  ];
-});
-
 // src/providers/transport/provider-model-discovery.ts
 async function discoverProviderProbeModel(providerName, displayName, exclude) {
   const def = getProviderByName(providerName);
@@ -31123,7 +31174,7 @@ async function discoverProviderProbeModel(providerName, displayName, exclude) {
       reason: failure ? `${displayName}: ${describeDiscoveryFailure(failure)}` : `${displayName} listed no models at ${def.modelDiscovery.path} \u2014 check the API key and that the subscription is active`
     };
   }
-  const ranked = rankDiscoveredModels(discovered).map((model) => model.id).filter(isChatCapable);
+  const ranked = rankDiscoveredModels(discovered).filter((model) => isReportedChatCapable(model.id, model.reported)).map((model) => model.id);
   if (ranked.length === 0) {
     return {
       model: null,
@@ -32945,11 +32996,14 @@ function sanitizeErrorMessage(message, maxLength = MAX_ERROR_MESSAGE_LENGTH) {
     return flattened;
   return `${flattened.slice(0, maxLength - 1).trimEnd()}\u2026`;
 }
-function wrapAnthropicError(status, message, errorType, upstreamStatus) {
+function wrapAnthropicError(status, message, errorType, upstreamStatus, providerMessage) {
   const type = errorType || statusToErrorType(status);
   const error = { type, message: sanitizeErrorMessage(message) };
   if (upstreamStatus !== undefined)
     error.upstream_status = upstreamStatus;
+  const oneLine = (providerMessage ?? "").replace(/\s+/g, " ").trim();
+  if (oneLine)
+    error.provider_message = sanitizeErrorMessage(oneLine);
   return { type: "error", error };
 }
 function extractUpstreamStatus(body) {
@@ -32963,11 +33017,38 @@ function extractUpstreamStatus(body) {
     return;
   }
 }
+function sseDataPayload(body) {
+  if (!/^\s*(event|data):/.test(body))
+    return;
+  const lines = body.split(/\r?\n/).filter((line) => line.startsWith("data:"));
+  if (lines.length > 0) {
+    const joined = lines.map((line) => line.slice("data:".length).trim()).join("");
+    if (joined.length > 0)
+      return joined;
+  }
+  const at = body.search(/(^|\s)data:/);
+  if (at === -1)
+    return;
+  const payload = body.slice(body.indexOf("data:", at) + "data:".length).trim();
+  return payload.length > 0 ? payload : undefined;
+}
 function extractProviderMessage(body) {
   if (body == null)
     return "";
-  if (typeof body === "string")
+  if (typeof body === "string") {
+    const payload = sseDataPayload(body);
+    if (payload) {
+      try {
+        const inner = extractProviderMessage(JSON.parse(payload));
+        if (inner)
+          return inner;
+      } catch {
+        return payload;
+      }
+      return payload;
+    }
     return body;
+  }
   const candidates = [
     body?.error?.message,
     body?.message,
@@ -32977,7 +33058,7 @@ function extractProviderMessage(body) {
   ];
   for (const c of candidates) {
     if (typeof c === "string" && c.length > 0)
-      return c;
+      return extractProviderMessage(c);
     if (Array.isArray(c)) {
       const first = c.find((e) => typeof e?.msg === "string" && e.msg.length > 0);
       if (first)
@@ -33567,7 +33648,7 @@ var init_stream_head_sniffer = __esm(() => {
 });
 
 // src/handlers/shared/stream-parsers/anthropic-sse.ts
-function sseDataPayload(line) {
+function sseDataPayload2(line) {
   if (!line.startsWith("data:"))
     return null;
   const rest = line.slice(5);
@@ -33805,7 +33886,7 @@ data: {"type":"ping"}
           buffer = lines.pop() || "";
           for (const line of lines) {
             totalLines++;
-            const payload = sseDataPayload(line);
+            const payload = sseDataPayload2(line);
             if (filterThinking && payload !== null) {
               try {
                 const data = JSON.parse(payload);
@@ -35918,7 +35999,8 @@ class ComposedHandler {
           parsedErrorBody = undefined;
         }
         const providerMsg = extractProviderMessage(parsedErrorBody ?? errorText);
-        const msgTail = providerMsg ? ` (${providerMsg.length > 200 ? `${providerMsg.slice(0, 200)}\u2026` : providerMsg})` : "";
+        const oneLineMsg = providerMsg.replace(/\s+/g, " ").trim();
+        const msgTail = oneLineMsg ? ` (${oneLineMsg.length > 200 ? `${oneLineMsg.slice(0, 200)}\u2026` : oneLineMsg})` : "";
         logStderr(`Error [${this.provider.displayName}]: HTTP ${response.status}. ${hint}${msgTail}`);
         let providerErrorType;
         try {
@@ -35971,7 +36053,7 @@ class ComposedHandler {
             providerMessage: providerMsg,
             leadPhrase: isContextOverflowError(response.status, errorText) ? CONTEXT_OVERFLOW_PHRASE : undefined
           });
-          return c.json(wrapAnthropicError(400, surfaced, "invalid_request_error", response.status), 400);
+          return c.json(wrapAnthropicError(400, surfaced, "invalid_request_error", response.status, providerMsg), 400);
         }
         return c.json(ensureAnthropicErrorFormat(response.status, errorBody), response.status);
       }
@@ -36004,7 +36086,7 @@ class ComposedHandler {
             invocation_mode: this.options.invocationMode ?? "auto-route"
           });
         } catch {}
-        return c.json(wrapAnthropicError(503, surfaced, "overloaded_error"), 503);
+        return c.json(wrapAnthropicError(503, surfaced, "overloaded_error", undefined, settled.message), 503);
       }
       response = settled.response;
     }
@@ -36046,7 +36128,7 @@ class ComposedHandler {
             invocation_mode: this.options.invocationMode ?? "auto-route"
           });
         } catch {}
-        return isTerminal ? c.json(wrapAnthropicError(400, surfaced, "invalid_request_error"), 400) : c.json(wrapAnthropicError(503, surfaced, "overloaded_error"), 503);
+        return isTerminal ? c.json(wrapAnthropicError(400, surfaced, "invalid_request_error", undefined, settled.message), 400) : c.json(wrapAnthropicError(503, surfaced, "overloaded_error", undefined, settled.message), 503);
       }
       response = settled.response;
     }
@@ -38551,7 +38633,7 @@ var init_provider_definitions = __esm(() => {
       description: "Alibaba Model Studio PAYG (qpay@)"
     },
     {
-      createHandler: openaiHandler,
+      createHandler: noHandler("virtual", "No baseUrl. Exists only so nativeModelPatterns can steer a bare qwen* name; qwen-payg serves dashscope."),
       tier: "native",
       name: "qwen",
       displayName: "Qwen",
@@ -38881,6 +38963,7 @@ async function refreshCatalog(timeoutMs, options = {}) {
     catalogGenerationId: generationId
   };
   writeAllModelsCache(cache, options.cachePath);
+  _clearChatCapabilityIndex();
   _memCache = entries;
   _warmPromise = Promise.resolve();
   return {
@@ -38957,6 +39040,7 @@ var DEFAULT_CATALOG_URL = "https://us-central1-claudish-6da10.cloudfunctions.net
 var init_catalog_client = __esm(() => {
   init_all_models_cache();
   init_catalog_route_bindings();
+  init_probe_discovery();
 });
 
 // src/config-schema.ts
@@ -40094,12 +40178,16 @@ function modelsCatalogHas(ids, wireId) {
   const needle = wireId.trim().toLowerCase();
   return ids.some((id) => id.trim().toLowerCase() === needle);
 }
+function resolveAgainstModelsCatalog(provider, wireId, models) {
+  const entries = models.map(({ id, ...rest }) => ({ wireId: id, ...rest }));
+  return expandSelection(provider, wireId, entries);
+}
 async function providerServesModel(provider, wireId) {
   const def = getProviderByName(provider);
   if (def?.modelDiscovery) {
     const models = await discoverProviderModels(provider);
     if (models.length > 0) {
-      return modelsCatalogHas(models.map((m) => m.id), wireId) ? "serves" : "not-served";
+      return modelsCatalogHas(models.map((m) => m.id), resolveAgainstModelsCatalog(provider, wireId, models)) ? "serves" : "not-served";
     }
     getDiscoveryFailure(provider);
     return "unknown";
@@ -40117,6 +40205,7 @@ var init_model_availability = __esm(() => {
   init_catalog_route_bindings();
   init_catalog_client();
   init_model_discovery();
+  init_registry2();
   init_provider_definitions();
 });
 
@@ -44954,17 +45043,32 @@ function truncateKeepingLink(text, max = 400) {
   const prose = (lastSpace > room * 0.5 ? head.slice(0, lastSpace) : head).trimEnd();
   return `${prose}... ${url}`;
 }
+function oneLine4(text) {
+  return text.replace(/\s+/g, " ").trim();
+}
+function messageFromJson(text) {
+  try {
+    const parsed = JSON.parse(text);
+    const msg = parsed?.error?.provider_message || parsed?.error?.message || parsed?.error?.error?.message || parsed?.message || parsed?.detail;
+    return typeof msg === "string" && msg.length > 0 ? msg : undefined;
+  } catch {
+    return;
+  }
+}
+function unwrapFrame(text) {
+  const payload = sseDataPayload(text);
+  if (!payload)
+    return text;
+  return messageFromJson(payload) ?? payload;
+}
 function extractErrorMessage(body) {
   if (!body)
     return;
-  try {
-    const parsed = JSON.parse(body);
-    const msg = parsed?.error?.message || parsed?.error?.error?.message || parsed?.message || parsed?.detail;
-    if (typeof msg === "string" && msg.length > 0) {
-      return truncateKeepingLink(msg);
-    }
-  } catch {}
-  const trimmed = body.trim();
+  const direct = messageFromJson(body);
+  if (direct)
+    return truncateKeepingLink(oneLine4(unwrapFrame(direct)));
+  const unwrapped = unwrapFrame(body);
+  const trimmed = oneLine4(unwrapped);
   if (!trimmed)
     return;
   return truncateKeepingLink(trimmed);
@@ -45141,7 +45245,10 @@ function isContentEvent(parsed, eventType) {
   return false;
 }
 function withDetail(base, message) {
-  return message ? `${base} \u2014 ${message}` : base;
+  return message ? `${base} \u2014 ${stripRedundantHead(message)}` : base;
+}
+function stripRedundantHead(message) {
+  return message.replace(/^.{1,40}? error \(HTTP \d{3}\): /, "");
 }
 function describeProbeState(result) {
   const status = result.httpStatus ?? "";
@@ -59306,7 +59413,7 @@ function logModalityGapOnce(kept) {
   if (suspect.length === 0)
     return;
   _modalityGapLogged = true;
-  log(`[Models] ${suspect.length} row(s) may not be chat models and cannot be classified from the id: ` + `${suspect.slice(0, 8).map((m) => m.id).join(", ")}. The slim catalog carries no modality field \u2014 models-index gap, not a CLI regex.`);
+  log(`[Models] ${suspect.length} row(s) the catalog publishes as chat, from a generator family: ` + `${suspect.slice(0, 8).map((m) => m.id).join(", ")}. If one is not a chat model, its models-index row is wrong.`);
 }
 async function getFreeModels() {
   return [];
@@ -59762,13 +59869,13 @@ async function buildDiscoveredModelOutcome(provider, displayName, catalog) {
   }
   const served = rankDiscoveredModels(modelsCatalog.models);
   const servedCount = served.length;
-  const discovered = served.filter((m) => isChatCapable(m.id));
+  const discovered = served.filter((m) => isReportedChatCapable(m.id, m.reported));
   const chatCount = discovered.length;
   if (chatCount === 0) {
     return {
       kind: "all-filtered",
       servedCount,
-      sampleIds: served.filter((m) => !isChatCapable(m.id)).slice(0, 3).map((m) => m.id),
+      sampleIds: served.filter((m) => !isReportedChatCapable(m.id, m.reported)).slice(0, 3).map((m) => m.id),
       fallbackRows
     };
   }
@@ -60848,14 +60955,15 @@ function useProbeStore(store) {
 }
 function deriveLayout(width) {
   if (width < 60) {
-    return { barWidth: 0, tokWidth: 0, showBreakdown: false, pillFallback: true };
+    return { barWidth: 0, tokWidth: 0, showBreakdown: false, pillFallback: true, width };
   }
   if (width < 80) {
     return {
       barWidth: TIMELINE_BAR_NARROW,
       tokWidth: 0,
       showBreakdown: false,
-      pillFallback: false
+      pillFallback: false,
+      width
     };
   }
   if (width < 100) {
@@ -60863,14 +60971,16 @@ function deriveLayout(width) {
       barWidth: TIMELINE_BAR_FULL,
       tokWidth: 0,
       showBreakdown: true,
-      pillFallback: false
+      pillFallback: false,
+      width
     };
   }
   return {
     barWidth: TIMELINE_BAR_FULL,
     tokWidth: TOK_BAR_FULL,
     showBreakdown: true,
-    pillFallback: false
+    pillFallback: false,
+    width
   };
 }
 function computeRowWidth(layout, maxNameLen) {
@@ -60912,8 +61022,12 @@ function padEndSafe(s, n) {
     return s.slice(0, n);
   return s + " ".repeat(n - s.length);
 }
+function clipReason(s, width) {
+  const room = Math.max(8, width - 2);
+  return s.length <= room ? s : `${s.slice(0, room - 1)}\u2026`;
+}
 function stripAnsi3(text) {
-  return text.replace(/\x1b\[[0-9;]*[A-Za-z]/g, "");
+  return text.replace(/\x1b\[[0-9;]*[A-Za-z]/g, "").replace(/\s+/g, " ").trim();
 }
 function ishGreen() {
   return getThemeMode() === "light" ? "#047857" : "#00ff7f";
@@ -61031,7 +61145,19 @@ function ProgressBar({
     return /* @__PURE__ */ jsxs9("text", {
       children: [
         prefix,
-        renderNonLiveStatus(link, false)
+        renderNonLiveStatus(link, false, layout.width)
+      ]
+    });
+  }
+  if (link.status === "failed") {
+    const used = ELAPSED_COL + maxNameLen + 2;
+    return /* @__PURE__ */ jsxs9("text", {
+      children: [
+        prefix,
+        /* @__PURE__ */ jsx12("span", {
+          fg: C.red,
+          children: clipReason(`\u2717 ${stripAnsi3(link.error || "failed")}`, layout.width - used)
+        })
       ]
     });
   }
@@ -61044,7 +61170,7 @@ function ProgressBar({
           fg: C.dim,
           children: "  "
         }),
-        renderNonLiveStatus(link, true)
+        renderNonLiveStatus(link, true, layout.width)
       ]
     });
   }
@@ -61178,7 +61304,7 @@ function renderTimelineSlot(link, animFrame, barWidth) {
       });
   }
 }
-function renderNonLiveStatus(link, hasSlot) {
+function renderNonLiveStatus(link, hasSlot, width) {
   switch (link.status) {
     case "probing": {
       const elapsedMs = link.startTime ? Date.now() - link.startTime : 0;
@@ -61193,7 +61319,7 @@ function renderNonLiveStatus(link, hasSlot) {
         children: "\u2717"
       }) : /* @__PURE__ */ jsx12("span", {
         fg: C.red,
-        children: `\u2717 ${stripAnsi3(link.error || "failed")}`
+        children: clipReason(`\u2717 ${stripAnsi3(link.error || "failed")}`, width)
       });
     default:
       return /* @__PURE__ */ jsx12("span", {
@@ -61396,6 +61522,7 @@ function DetailLinkRow({
     ]
   });
   if (!isLive || !probe?.timing) {
+    const used = 2 + 1 + 1 + provW + 2 + 3;
     return /* @__PURE__ */ jsxs9("text", {
       children: [
         lead,
@@ -61405,7 +61532,7 @@ function DetailLinkRow({
         }),
         /* @__PURE__ */ jsx12("span", {
           fg: C.red,
-          children: shortFailureReason(probe, link.hasCredentials)
+          children: clipReason(shortFailureReason(probe, link.hasCredentials), layout.width - used)
         })
       ]
     });
@@ -62141,7 +62268,7 @@ function ProbeApp({
     ]
   }, state.phase);
 }
-var ANIM_FRAMES2, TIMELINE_BAR_FULL = 24, TIMELINE_BAR_NARROW = 12, TOK_BAR_FULL = 14, TOTAL_COL = 7, STAGE_NUM_W2 = 6, BREAKDOWN_COL, TOK_VALUE_COL = 7, TRACK_CHAR = "\xB7", BAR_FILL = "\u2588", BANNER_ROWS = 7, SCROLL_HINT_ROWS = 1, LEGEND_ROWS = 2, MIN_LIST_H = 4, TAB_BAR_ROWS = 2;
+var ANIM_FRAMES2, TIMELINE_BAR_FULL = 24, TIMELINE_BAR_NARROW = 12, TOK_BAR_FULL = 14, TOTAL_COL = 7, ELAPSED_COL = 11, STAGE_NUM_W2 = 6, BREAKDOWN_COL, TOK_VALUE_COL = 7, TRACK_CHAR = "\xB7", BAR_FILL = "\u2588", BANNER_ROWS = 7, SCROLL_HINT_ROWS = 1, LEGEND_ROWS = 2, MIN_LIST_H = 4, TAB_BAR_ROWS = 2;
 var init_probe_tui_app = __esm(() => {
   init_probe_live();
   init_theme_mode();
@@ -63665,6 +63792,7 @@ async function probeModelRouting(models, jsonOutput, options = { live: true, tim
     activeTab: "summary"
   };
   const tui = await startProbeTui(initialState);
+  setStderrQuiet(true);
   const addStep = (name, status) => {
     tui.store.setState((prev) => ({
       ...prev,
@@ -63852,6 +63980,7 @@ async function probeModelRouting(models, jsonOutput, options = { live: true, tim
       } catch {}
     }
     await tui.shutdown();
+    setStderrQuiet(false);
   }
 }
 function printHelp2() {
@@ -64279,6 +64408,7 @@ var __filename3, __dirname3;
 var init_cli = __esm(() => {
   init_base_api_format();
   init_config2();
+  init_logger();
   init_model_loader();
   init_model_selector();
   init_probe_results_printer();