@jacobbd/relay-ai 0.6.1 → 0.6.3

package/README.md

@@ -17,9 +17,9 @@
 
 📺 **Watch the Demos**
 
-| **Claude Code / Cowork / Desktop** | **Codex CLI & Desktop App** | **Gemini CLI** | **v0.4.1: UI & Antigravity** |
-|:---:|:---:|:---:|:---:|
-| [![Claude Demo](https://img.youtube.com/vi/IvsUPHLhX0o/mqdefault.jpg)](https://youtu.be/IvsUPHLhX0o) | [![Codex Demo](https://img.youtube.com/vi/42oiOB8IAu4/mqdefault.jpg)](https://youtu.be/42oiOB8IAu4) | [![Gemini Demo](https://img.youtube.com/vi/g7JKvqOHJl4/mqdefault.jpg)](https://www.youtube.com/watch?v=g7JKvqOHJl4) | [![UI & Antigravity Demo](https://img.youtube.com/vi/8vXJ0LfpdoY/mqdefault.jpg)](https://www.youtube.com/watch?v=8vXJ0LfpdoY) |
+| **Claude Code / Cowork / Desktop** | **Codex CLI & Desktop App** | **Gemini CLI** | **v0.4.1: UI & Antigravity** | **Server Gateway** |
+|:---:|:---:|:---:|:---:|:---:|
+| [![Claude Demo](https://img.youtube.com/vi/IvsUPHLhX0o/mqdefault.jpg)](https://youtu.be/IvsUPHLhX0o) | [![Codex Demo](https://img.youtube.com/vi/42oiOB8IAu4/mqdefault.jpg)](https://youtu.be/42oiOB8IAu4) | [![Gemini Demo](https://img.youtube.com/vi/g7JKvqOHJl4/mqdefault.jpg)](https://www.youtube.com/watch?v=g7JKvqOHJl4) | [![UI & Antigravity Demo](https://img.youtube.com/vi/8vXJ0LfpdoY/mqdefault.jpg)](https://www.youtube.com/watch?v=8vXJ0LfpdoY) | [![Server Gateway Demo](https://img.youtube.com/vi/4CEw0MQUE9I/mqdefault.jpg)](https://www.youtube.com/watch?v=4CEw0MQUE9I) |
 
 **relay-ai** is an interactive CLI — and now a **visual launcher** — that connects AI coding tools to any provider and runs local API gateways on your machine. It supports **Claude Code**, **Claude Desktop (Cowork + Code)**, the **OpenAI Codex CLI**, the **ChatGPT desktop app in Codex mode (macOS + Windows)**, **Google Gemini CLI**, and experimental **Antigravity CLI / IDE** support.
 
@@ -455,8 +455,9 @@ Launch **Claude Desktop** (macOS or Windows) with registry providers:
 relay-ai claude-app
 ```
 
-This command automates the "Third-Party Inference" (Developer Mode) setup. It temporarily configures Claude Desktop to point at a local gateway, launches the app, and routes traffic to your chosen provider.
+This command automates the "Third-Party Inference" (Developer Mode) setup. It temporarily configures Claude Desktop to point at a local gateway, launches the app, and exposes the selected model followed by your available saved favorites (up to 20 models total).
 
+- **Selected + favorites:** The selected model is listed first; duplicates and unavailable favorites are skipped. Claude Desktop ultimately controls which discovered model it activates initially.
 - **Keep the terminal open:** The proxy runs in the foreground.
 - **Ctrl+C to restore:** When you're done, press `Ctrl+C` in the terminal to automatically restore Claude Desktop to its normal Anthropic cloud mode.
 - **Cleanup:** If the terminal crashes, run `relay-ai claude-app --restore`.

package/dist/{chunk-P4S42QJK.js → chunk-WHEMJ7C3.js}

@@ -11,7 +11,7 @@ import { join } from "path";
 // package.json
 var package_default = {
   name: "@jacobbd/relay-ai",
-  version: "0.6.1",
+  version: "0.6.3",
   publishConfig: {
     access: "public"
   },
@@ -46,7 +46,7 @@ var package_default = {
     node: ">=18"
   },
   scripts: {
-    build: "tsup && mkdir -p dist/ui/public && cp -r src/ui/public/. dist/ui/public/",
+    build: "tsup && node scripts/copy-ui-assets.mjs",
     dev: "tsup --watch",
     test: "vitest run",
     "test:watch": "vitest",
@@ -139,6 +139,14 @@ var CONFLICTING_ENV_VARS = [
   "ANTHROPIC_DEFAULT_SONNET_MODEL",
   "ANTHROPIC_DEFAULT_HAIKU_MODEL"
 ];
+var PARENT_SESSION_ENV_VARS = [
+  "CLAUDECODE",
+  "CLAUDE_CODE_CHILD_SESSION",
+  "CLAUDE_CODE_SESSION_ID",
+  "CLAUDE_CODE_HOST_SESSION_ID",
+  "CLAUDE_CODE_ENTRYPOINT",
+  "CLAUDE_PID"
+];
 var OPENCODE_CACHE_PATH = join(homedir(), ".cache", "opencode", "models.json");
 var MAX_MODEL_CATALOG = 20;
 var VERTEX_ANTHROPIC_NPM = "@ai-sdk/google-vertex/anthropic";
@@ -3045,6 +3053,9 @@ function buildChildEnv(baseUrl, model, apiKey, proxyPort, contextWindow, enableG
   for (const name of CONFLICTING_ENV_VARS) {
     delete env[name];
   }
+  for (const name of PARENT_SESSION_ENV_VARS) {
+    delete env[name];
+  }
   env["ANTHROPIC_BASE_URL"] = proxyPort ? `http://127.0.0.1:${proxyPort}` : baseUrl;
   env["ANTHROPIC_API_KEY"] = apiKey;
   const bareModel = stripOneMContextSuffix(model);
@@ -4243,7 +4254,7 @@ function maskGatewayModelId(aliasId) {
 // src/server/models.ts
 var CREATED_AT_ISO = "2025-01-01T00:00:00Z";
 var CREATED_AT_UNIX = 1735689600;
-function formatAnthropicModelEntry(id, displayName, contextWindow) {
+function formatAnthropicModelEntry(id, displayName, contextWindow, options) {
   const maxInput = resolveContextWindow(id, contextWindow);
   return {
     id,
@@ -4251,12 +4262,18 @@ function formatAnthropicModelEntry(id, displayName, contextWindow) {
     display_name: displayName,
     created_at: CREATED_AT_ISO,
     context_window: maxInput,
-    max_input_tokens: maxInput
+    max_input_tokens: maxInput,
+    ...options?.supportsOneM !== void 0 ? { supports_1m: options.supportsOneM } : {}
   };
 }
 function formatAnthropicModelList(entries) {
   return {
-    data: entries.map((entry) => formatAnthropicModelEntry(entry.id, entry.name, entry.contextWindow)),
+    data: entries.map((entry) => formatAnthropicModelEntry(
+      entry.id,
+      entry.name,
+      entry.contextWindow,
+      { supportsOneM: entry.supportsOneM }
+    )),
     has_more: false,
     first_id: entries[0]?.id ?? null,
     last_id: entries.at(-1)?.id ?? null
@@ -4271,31 +4288,65 @@ function gatewayProviderId(model) {
 function gatewayAliasId(model) {
   return aliasModelId(model.id, gatewayProviderId(model));
 }
+function openAiIdCollisions(models) {
+  const counts = /* @__PURE__ */ new Map();
+  for (const model of models) counts.set(model.id, (counts.get(model.id) ?? 0) + 1);
+  const collisions = /* @__PURE__ */ new Set();
+  for (const [id, count] of counts) if (count > 1) collisions.add(id);
+  return collisions;
+}
+function openAiExposedId(model, collisions) {
+  return collisions.has(model.id) ? `${gatewayProviderId(model)}/${model.id}` : model.id;
+}
 function exposedGatewayAliasId(model, opts) {
-  const alias = gatewayAliasId(model);
-  return opts?.maskGatewayIds ? maskGatewayModelId(alias) : alias;
+  const singleOneM = usesSingleOneMEntry(model, opts);
+  const alias = gatewayAliasId(singleOneM ? { ...model, id: stripOneMContextSuffix(model.id) } : model);
+  const exposed = opts?.maskGatewayIds ? maskGatewayModelId(alias) : alias;
+  return singleOneM ? `${stripOneMContextSuffix(exposed)}[1m]` : exposed;
 }
 function gatewayDisplayName(model, opts) {
-  if (!opts?.maskGatewayIds) return model.name;
-  return `${model.name} (${gatewayProviderLabel(model)})`;
+  const name = opts?.maskGatewayIds ? `${model.name} (${gatewayProviderLabel(model)})` : model.name;
+  return usesSingleOneMEntry(model, opts) && !/\b1m$/i.test(name) ? `${name} 1M` : name;
 }
 function formatGatewayAnthropicModels(models, opts) {
   return formatAnthropicModelList(
     models.map((model) => ({
       id: exposedGatewayAliasId(model, opts),
       name: gatewayDisplayName(model, opts),
-      contextWindow: model.contextWindow
+      contextWindow: model.contextWindow,
+      supportsOneM: usesSingleOneMEntry(model, opts) ? false : void 0
     }))
   );
 }
+function usesSingleOneMEntry(model, opts) {
+  return opts?.longContextDisplay === "single-1m" && resolveContextWindow(stripOneMContextSuffix(model.id), model.contextWindow) >= 1e6;
+}
 function createGatewayModelCatalog(models, opts) {
   const byId = /* @__PURE__ */ new Map();
+  const collisions = openAiIdCollisions(models);
   for (const model of models) {
-    byId.set(model.id, model);
+    if (!byId.has(model.id)) byId.set(model.id, model);
+    const scopedId = openAiExposedId(model, collisions);
+    if (scopedId !== model.id) byId.set(scopedId, model);
     const alias = exposedGatewayAliasId(model, opts);
     if (alias !== model.id) byId.set(alias, model);
+    const singleOneM = usesSingleOneMEntry(model, opts);
+    if (singleOneM) {
+      const bareModel = { ...model, id: stripOneMContextSuffix(model.id) };
+      const rawBareAlias = gatewayAliasId(bareModel);
+      const exposedBareAlias = opts?.maskGatewayIds ? maskGatewayModelId(rawBareAlias) : rawBareAlias;
+      for (const compatibleId of [
+        stripOneMContextSuffix(model.id),
+        rawBareAlias,
+        `${rawBareAlias}[1m]`,
+        exposedBareAlias,
+        `${exposedBareAlias}[1m]`
+      ]) {
+        byId.set(compatibleId, model);
+      }
+    }
     if (opts?.maskGatewayIds) {
-      const rawAlias = gatewayAliasId(model);
+      const rawAlias = gatewayAliasId(singleOneM ? { ...model, id: stripOneMContextSuffix(model.id) } : model);
       if (rawAlias !== alias) byId.set(rawAlias, model);
     }
   }
@@ -4308,14 +4359,14 @@ function upstreamModelId(model) {
   const id = model.upstreamModelId ?? model.id;
   return id.replace(/\[1m\]$/i, "");
 }
-function buildDedupedModelRows(models, opts) {
+function buildDedupedModelRows(models, opts, collisions = openAiIdCollisions(models)) {
   const seen = /* @__PURE__ */ new Set();
   const rows = [];
   for (const model of [...models].sort((a, b) => a.name.localeCompare(b.name))) {
     const row = {
       name: model.name,
       anthropicId: exposedGatewayAliasId(model, opts),
-      openaiId: model.id
+      openaiId: openAiExposedId(model, collisions)
     };
     const key = `${row.name}\0${row.anthropicId}\0${row.openaiId}`;
     if (seen.has(key)) continue;
@@ -4328,10 +4379,11 @@ function supportsDirectOpenAIChatCompletions(model) {
   return model.modelFormat === "openai" && (!!model.completionsUrl || model.sourceBackend === "zen" || model.sourceBackend === "go");
 }
 function formatOpenAIModels(models) {
+  const collisions = openAiIdCollisions(models);
   return {
     object: "list",
     data: models.map((model) => ({
-      id: model.id,
+      id: openAiExposedId(model, collisions),
       object: "model",
       created: CREATED_AT_UNIX,
       owned_by: model.sourceBackend
@@ -8721,8 +8773,9 @@ function translateOpenAiRequest(body) {
         break;
       case "assistant": {
         const parts = [];
-        if (typeof msg.content === "string" && msg.content) {
-          parts.push({ type: "text", text: msg.content });
+        const assistantText = typeof msg.content === "string" ? msg.content : Array.isArray(msg.content) ? msg.content.filter((p8) => p8?.type === "text" && typeof p8.text === "string").map((p8) => p8.text).join("") : "";
+        if (assistantText) {
+          parts.push({ type: "text", text: assistantText });
         }
         for (const tc of msg.tool_calls ?? []) {
           parts.push({
@@ -8783,14 +8836,31 @@ function translateOpenAiRequest(body) {
     maxOutputTokens: body.max_completion_tokens ?? body.max_tokens
   };
 }
+function toOpenAiFinishReason(reason) {
+  switch (reason) {
+    case "tool-calls":
+      return "tool_calls";
+    case "content-filter":
+      return "content_filter";
+    case "length":
+      return "length";
+    case "stop":
+      return "stop";
+    default:
+      return "stop";
+  }
+}
 async function generateOpenAiResponse(model, params, responseModelId) {
   const result = await generateText2({ model, ...params });
   const message = { role: "assistant", content: result.text || null };
+  if (result.reasoningText || result.reasoning) {
+    message.reasoning_content = result.reasoningText ?? result.reasoning;
+  }
   if (result.toolCalls?.length) {
     message.tool_calls = result.toolCalls.map((tc) => ({
       id: tc.toolCallId,
       type: "function",
-      function: { name: tc.toolName, arguments: JSON.stringify(tc.args) }
+      function: { name: tc.toolName, arguments: JSON.stringify(tc.input ?? tc.args ?? {}) }
     }));
   }
   return {
@@ -8798,7 +8868,7 @@ async function generateOpenAiResponse(model, params, responseModelId) {
     object: "chat.completion",
     created: Math.floor(Date.now() / 1e3),
     model: responseModelId,
-    choices: [{ index: 0, message, finish_reason: result.finishReason || "stop" }],
+    choices: [{ index: 0, message, finish_reason: toOpenAiFinishReason(result.finishReason) }],
     usage: {
       prompt_tokens: result.usage?.promptTokens ?? 0,
       completion_tokens: result.usage?.completionTokens ?? 0,
@@ -8806,7 +8876,7 @@ async function generateOpenAiResponse(model, params, responseModelId) {
     }
   };
 }
-async function streamOpenAiResponse(model, params, responseModelId, onChunk) {
+async function streamOpenAiResponse(model, params, responseModelId, onChunk, log7) {
   const { fullStream } = streamText2({ model, ...params });
   const baseData = {
     id: `chatcmpl-${Date.now()}`,
@@ -8817,23 +8887,68 @@ async function streamOpenAiResponse(model, params, responseModelId, onChunk) {
   const send = (delta, finish_reason = null) => onChunk(`data: ${JSON.stringify({ ...baseData, choices: [{ index: 0, delta, finish_reason }] })}
 
 `);
+  const streamedToolIndex = /* @__PURE__ */ new Map();
+  let nextToolIndex = 0;
+  const seenPartTypes = /* @__PURE__ */ new Set();
+  let toolCallChunksEmitted = 0;
   for await (const part of fullStream) {
     const p8 = part;
+    seenPartTypes.add(p8.type);
     switch (p8.type) {
       case "text-delta":
         send({ role: "assistant", content: p8.textDelta ?? p8.text ?? "" });
         break;
+      case "reasoning-delta":
+        send({ role: "assistant", reasoning_content: p8.text ?? p8.delta ?? "" });
+        break;
       case "tool-input-start":
-      case "tool-call-streaming-start":
-        send({ role: "assistant", tool_calls: [{ index: 0, id: p8.id ?? p8.toolCallId, type: "function", function: { name: p8.toolName, arguments: "" } }] });
+      case "tool-call-streaming-start": {
+        const id = p8.id ?? p8.toolCallId ?? "";
+        const index = nextToolIndex++;
+        streamedToolIndex.set(id, index);
+        send({ role: "assistant", tool_calls: [{ index, id, type: "function", function: { name: p8.toolName, arguments: "" } }] });
+        toolCallChunksEmitted++;
         break;
+      }
       case "tool-input-delta":
-      case "tool-call-delta":
-        send({ tool_calls: [{ index: 0, function: { arguments: p8.delta ?? p8.text ?? p8.argsTextDelta ?? "" } }] });
+      case "tool-call-delta": {
+        const id = p8.id ?? p8.toolCallId ?? "";
+        const index = streamedToolIndex.get(id) ?? 0;
+        send({ tool_calls: [{ index, function: { arguments: p8.delta ?? p8.text ?? p8.argsTextDelta ?? "" } }] });
+        break;
+      }
+      case "tool-call": {
+        const id = p8.toolCallId ?? "";
+        if (streamedToolIndex.has(id)) break;
+        const index = nextToolIndex++;
+        send({
+          role: "assistant",
+          tool_calls: [{ index, id, type: "function", function: { name: p8.toolName, arguments: JSON.stringify(p8.input ?? {}) } }]
+        });
+        toolCallChunksEmitted++;
         break;
+      }
       case "finish":
-        send({}, p8.finishReason || "stop");
+        log7?.(() => `openai stream parts=[${[...seenPartTypes].join(",")}] toolCallChunks=${toolCallChunksEmitted} finishReason=${p8.finishReason}`);
+        send({}, toOpenAiFinishReason(p8.finishReason));
         break;
+      case "error": {
+        const errMsg = typeof p8.error === "string" ? p8.error : formatUpstreamError(p8.error);
+        log7?.(() => `openai stream error parts=[${[...seenPartTypes].join(",")}]: ${errMsg}`);
+        log7?.(() => {
+          try {
+            return `openai stream error raw: ${JSON.stringify(p8.error, Object.getOwnPropertyNames(p8.error ?? {})).slice(0, 3e3)}`;
+          } catch {
+            return `openai stream error raw: (unserializable) ${String(p8.error)}`;
+          }
+        });
+        send({ role: "assistant", content: `
+
+[relay-ai upstream error: ${errMsg}]` });
+        send({}, "stop");
+        onChunk("data: [DONE]\n\n");
+        return;
+      }
     }
   }
   onChunk("data: [DONE]\n\n");
@@ -9023,6 +9138,25 @@ async function handleOpenAIChatCompletions(req, res, options, modelCache, plog)
     sendJson(res, 400, { error: { message: "Invalid JSON body" } });
     return;
   }
+  plog(() => `openai-chat-completions raw body: ${JSON.stringify(body).slice(0, 6e3)}`);
+  plog(() => {
+    const messages = Array.isArray(body.messages) ? body.messages : [];
+    const summary = messages.map((m, i) => {
+      const msg = m;
+      const contentType = msg.content === null ? "null" : Array.isArray(msg.content) ? "array" : typeof msg.content;
+      let partShapes = "";
+      if (Array.isArray(msg.content)) {
+        partShapes = " parts=[" + msg.content.map((p8) => {
+          const part = p8;
+          const keys = Object.keys(part).join(",");
+          const textLen = typeof part.text === "string" ? part.text.length : void 0;
+          return `{type=${part.type} keys=${keys}${textLen !== void 0 ? ` textLen=${textLen}` : ""}}`;
+        }).join(",") + "]";
+      }
+      return `#${i} role=${msg.role} hasContent=${"content" in msg} contentType=${contentType} toolCalls=${Array.isArray(msg.tool_calls) ? msg.tool_calls.length : 0}${partShapes}`;
+    });
+    return `openai-chat-completions message shapes: [${summary.join(" | ")}]`;
+  });
   const model = lookupModel(res, options.catalog, body.model);
   if (!model) return;
   if (supportsDirectOpenAIChatCompletions(model)) {
@@ -9032,7 +9166,9 @@ async function handleOpenAIChatCompletions(req, res, options, modelCache, plog)
     }
     const completionsUrl = model.completionsUrl ? model.completionsUrl : `${backendFor(options, model).baseUrl}/v1/chat/completions`;
     const apiKey2 = model.apiKey ?? options.apiKey;
-    await relayAnthropicMessages(res, completionsUrl, body, apiKey2, Boolean(body.stream));
+    const forwardBody = { ...body, model: upstreamModelId(model) };
+    plog(() => `openai-direct-passthrough \u2192 ${completionsUrl} model=${forwardBody.model} stream=${Boolean(body.stream)}`);
+    await relayAnthropicMessages(res, completionsUrl, forwardBody, apiKey2, Boolean(body.stream), void 0, void 0, (message) => plog(message));
     return;
   }
   const npm = model.npm || (model.modelFormat === "anthropic" ? "@ai-sdk/anthropic" : void 0);
@@ -9054,7 +9190,7 @@ async function handleOpenAIChatCompletions(req, res, options, modelCache, plog)
         "Cache-Control": "no-cache",
         "Connection": "keep-alive"
       });
-      await streamOpenAiResponse(languageModel, params, responseModelId, (chunk) => res.write(chunk));
+      await streamOpenAiResponse(languageModel, params, responseModelId, (chunk) => res.write(chunk), plog);
       res.end();
     } else {
       const response = await generateOpenAiResponse(languageModel, params, responseModelId);
@@ -9395,10 +9531,11 @@ function formatModelCatalogLines(models, gateway) {
     }
     list.push(model);
   }
+  const collisions = openAiIdCollisions(models);
   const lines = ["Model catalog:", ""];
   const sortedGroups = [...groups.entries()].sort((a, b) => a[0].localeCompare(b[0]));
   for (const [label, groupModels] of sortedGroups) {
-    const rows = buildDedupedModelRows(groupModels, gateway);
+    const rows = buildDedupedModelRows(groupModels, gateway, collisions);
     const hiddenDuplicates = groupModels.length - rows.length;
     const duplicateNote = hiddenDuplicates > 0 ? `, ${hiddenDuplicates} duplicate${hiddenDuplicates !== 1 ? "s" : ""} hidden` : "";
     const nameWidth = cappedWidth(rows.map((row) => row.name), "Model", 28);
@@ -10580,6 +10717,8 @@ function winLocalAppData2() {
 function winClaudeExeCandidates() {
   const local = winLocalAppData2();
   const bases = [
+    // Squirrel install folder used by the Anthropic Claude desktop installer.
+    join12(local, "AnthropicClaude"),
     join12(local, "Programs", "Claude"),
     join12(local, "Claude")
   ];
@@ -10877,6 +11016,7 @@ export {
   appendCodexBodyDump,
   getGeminiProxyDebugLogPath,
   getUiDebugLogPath,
+  getServerDebugLogPath,
   makeTraceLogger,
   writeSecureLogLine,
   printTraceLog,
@@ -10892,6 +11032,7 @@ export {
   extractApiKey,
   sendJson,
   gatewayProviderLabel,
+  openAiIdCollisions,
   createGatewayModelCatalog,
   buildDedupedModelRows,
   grabRoundTripSignature,
@@ -10962,4 +11103,4 @@ export {
   supportsClaudeTransparentMode,
   buildHttpProxyRoutes
 };
-//# sourceMappingURL=chunk-P4S42QJK.js.map
+//# sourceMappingURL=chunk-WHEMJ7C3.js.map