[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_170":171,"_172":173},"routes/_app.x.$handle.$name",{"_99":98,"_37":48,"_107":89,"_32":108,"_13":14,"_109":-5,"_11":110,"_111":112,"_113":114,"_115":116,"_117":118,"_26":-5,"_119":-5,"_120":-5,"_121":30,"_122":123,"_124":125,"_126":-5,"_127":-5,"_128":129},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":18,"_19":-5},"project",{"_46":47,"_37":48,"_41":49,"_50":-5,"_51":30,"_52":53,"_54":53,"_55":56,"_57":-5,"_19":58,"_59":60,"_61":30,"_62":-5,"_63":-5,"_64":65,"_66":67,"_68":-5,"_69":-5,"_70":-5,"_71":-5,"_72":73,"_74":73,"_75":76,"_77":76,"_78":76,"_79":73,"_80":73,"_81":82,"_83":84,"_85":86,"_87":30},"version",8,"changes",[25],"aiChatInfo",{"_20":21,"_22":-5,"_23":-5,"_24":-5},"userId","aiChatMessageId","6f05a33a-351e-4ef6-a469-5df64976113c","aiChatThreadId","aiChatThreadUserId","aiChatThreadTitle",{"_26":27,"_28":-7,"_29":30,"_31":30,"_32":33,"_34":33},"file",{"_35":36,"_37":33,"_38":26,"_39":40,"_41":42,"_43":14,"_44":45,"_29":30},"previous","deleted",false,"deletedHere","path","README.md","latestPath","fileId","2fa931a6-3103-11f1-ba5c-42dde27851f2","name","fileType","content","# OpenAI-Compatible API Proxy v2.0\n\nA production-ready, non-streaming backend that exposes OpenAI-compatible endpoints for both **OpenAI** and **Anthropic Claude** models — including full **tool_calls / function calling** support.\n\n## Base URL\n\n```\nhttps://tmp100nuaa--b8479a5a310011f1a25642dde27851f2.web.val.run/api/v1\n```\n\n---\n\n## Setup\n\n### For Claude models (required)\n\nAdd your Anthropic API key as an environment variable in this val:\n\n| Key | Value |\n|-----|-------|\n| `ANTHROPIC_API_KEY` | `sk-ant-...` |\n\nOpenAI models work out-of-the-box via Val Town's built-in proxy (no key needed).\n\n---\n\n## Endpoints\n\n### `GET /api/v1/models`\nReturns all available models in OpenAI format. No auth required.\n\n```bash\ncurl https://.../api/v1/models\n```\n\n### `POST /api/v1/chat/completions`\nNon-streaming chat completions with tool_calls support.\n\n**Auth:** Any `Bearer ` header is accepted.\n\n---\n\n## Model Reference\n\n### OpenAI models (built-in, no key needed)\n\n| Model ID | Description |\n|----------|-------------|\n| `gpt-4o` | GPT-4o |\n| `gpt-4o-mini` | GPT-4o Mini |\n| `gpt-4.1` / `gpt-4.1-mini` / `gpt-4.1-nano` | GPT-4.1 family |\n| `gpt-5-nano` | GPT-5 Nano |\n\n### Claude models (requires `ANTHROPIC_API_KEY`)\n\n| Model ID | Upstream | Description |\n|----------|----------|-------------|\n| `claude-opus-4` | `claude-opus-4-5` | Most capable |\n| `claude-sonnet-4` | `claude-sonnet-4-5` | Balanced |\n| `claude-3-5-sonnet` | `claude-3-5-sonnet-20241022` | Previous gen |\n| `claude-3-5-haiku` | `claude-3-5-haiku-20241022` | Fast & cheap |\n| `claude-haiku-3-5` | `claude-haiku-3-5-20241022` | Alias |\n| `claude-3-opus` | `claude-3-opus-20240229` | Legacy |\n\n### Friendly aliases (auto-mapped)\n\n| Alias | Routes to |\n|-------|-----------|\n| `gpt-5`, `gpt-5-turbo`, `gpt-4`, `gpt-4-turbo` | `gpt-4o` |\n| `claude-4.6-sonnet`, `claude-4-sonnet`, `claude-sonnet-4-5` | `claude-sonnet-4` |\n| `claude-4-opus`, `claude-4.6-opus` | `claude-opus-4` |\n| `claude-3-sonnet` | `claude-3-5-sonnet` |\n| `gemini-2.5-pro`, `gemini-1.5-pro` | `gpt-4o` (fallback) |\n\n---\n\n## Examples\n\n### Non-streaming chat (OpenAI)\n```bash\ncurl -X POST https://.../api/v1/chat/completions \\\n -H \"Authorization: Bearer any-token\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"gpt-4o-mini\",\n \"messages\": [{\"role\": \"user\", \"content\": \"Hello!\"}]\n }'\n```\n\n### Non-streaming chat (Claude)\n```bash\ncurl -X POST https://.../api/v1/chat/completions \\\n -H \"Authorization: Bearer any-token\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"claude-3-5-sonnet\",\n \"messages\": [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"What is 2+2?\"}\n ]\n }'\n```\n\n### Tool calling (Claude)\n```bash\ncurl -X POST https://.../api/v1/chat/completions \\\n -H \"Authorization: Bearer any-token\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"claude-3-5-sonnet\",\n \"messages\": [{\"role\": \"user\", \"content\": \"Whats the weather in Beijing?\"}],\n \"tools\": [{\n \"type\": \"function\",\n \"function\": {\n \"name\": \"get_weather\",\n \"description\": \"Get weather for a city\",\n \"parameters\": {\n \"type\": \"object\",\n \"properties\": {\n \"city\": {\"type\": \"string\"},\n \"unit\": {\"type\": \"string\", \"enum\": [\"celsius\", \"fahrenheit\"]}\n },\n \"required\": [\"city\"]\n }\n }\n }],\n \"tool_choice\": \"auto\"\n }'\n```\n\n### Multi-turn tool calling (returning tool result)\n```bash\ncurl -X POST https://.../api/v1/chat/completions \\\n -H \"Authorization: Bearer any-token\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"claude-3-5-sonnet\",\n \"messages\": [\n {\"role\": \"user\", \"content\": \"Whats the weather in Beijing?\"},\n {\"role\": \"assistant\", \"content\": null, \"tool_calls\": [{\"id\": \"call_123\", \"type\": \"function\", \"function\": {\"name\": \"get_weather\", \"arguments\": \"{\\\"city\\\":\\\"Beijing\\\"}\"}}]},\n {\"role\": \"tool\", \"tool_call_id\": \"call_123\", \"content\": \"{\\\"temp\\\": 22, \\\"condition\\\": \\\"Sunny\\\"}\"}\n ]\n }'\n```\n\n### Python (OpenAI SDK)\n```python\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=\"any-token\",\n base_url=\"https://tmp100nuaa--b8479a5a310011f1a25642dde27851f2.web.val.run/api/v1\"\n)\n\n# Simple chat\nresponse = client.chat.completions.create(\n model=\"claude-3-5-sonnet\",\n messages=[{\"role\": \"user\", \"content\": \"Hello Claude!\"}]\n)\nprint(response.choices[0].message.content)\n\n# With tools\ntools = [{\n \"type\": \"function\",\n \"function\": {\n \"name\": \"get_weather\",\n \"description\": \"Get current weather\",\n \"parameters\": {\n \"type\": \"object\",\n \"properties\": {\"city\": {\"type\": \"string\"}},\n \"required\": [\"city\"]\n }\n }\n}]\n\nresponse = client.chat.completions.create(\n model=\"claude-3-5-sonnet\",\n messages=[{\"role\": \"user\", \"content\": \"Weather in Shanghai?\"}],\n tools=tools,\n tool_choice=\"auto\"\n)\nmsg = response.choices[0].message\nif msg.tool_calls:\n print(\"Tool called:\", msg.tool_calls[0].function.name)\n print(\"Arguments:\", msg.tool_calls[0].function.arguments)\n```\n\n---\n\n## Notes\n\n- **Streaming is disabled** — set `stream: false` or omit it entirely.\n- **tool_choice mapping**: `\"auto\"` → Anthropic `auto`, `\"required\"` → Anthropic `any`, `{\"type\":\"function\",\"function\":{\"name\":\"...\"}}` → Anthropic `tool` type.\n- **system messages** are correctly lifted to Anthropic's top-level `system` parameter.\n- **tool role messages** (tool results) are converted to Anthropic's `tool_result` block format.\n","createdAt",["D",1775402517936],"branchVersion","branchId","9e5c7d86-3100-11f1-a256-42dde27851f2","id","9e5c250c-3100-11f1-a256-42dde27851f2","openai-compatible-proxy",["D",1775401415177],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","OpenAI-compatible API proxy using Val Town built-in AI models","imageUrl","019d5e2a-1b3f-7266-8f0e-ed31e2701dd4","pinnedFileId","b8479a5a-3100-11f1-a256-42dde27851f2","pinPreview","pinMigrated","screenshotVersion","likeCount",0,"orgId","019d5e2a-1b3f-7266-8f0e-f3afc8b1ccce","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_99":98,"_106":102},"org",{"_99":98,"_100":-5,"_101":102,"_19":58,"_103":104,"_105":30},"branch",{"_13":88,"_37":89,"_46":45,"_90":47,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":48,"_97":98},"isOwned",12,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","tmp100nuaa","handle","displayName","avatar","https://img.clerk.com/eyJ0eXBlIjoiZGVmYXVsdCIsImlpZCI6Imluc18yTm10cGhQUklwU0tqZHNiOHNnNGd2ZjJURFUiLCJyaWQiOiJ1c2VyXzNCd1hyRHZOdHcwNk9LZVl4dU1LcFJSWkZFUSJ9","tier","free","canRestrictVals","profileImageUrl","branchName","","adoptOrgId",{"_46":47,"_37":48,"_41":166,"_50":-5,"_51":30,"_52":53,"_54":53,"_55":56,"_57":-5,"_19":58,"_59":60,"_61":30,"_62":-5,"_63":-5,"_64":65,"_66":67,"_68":-5,"_69":-5,"_70":-5,"_71":-5,"_72":73,"_74":73,"_75":76,"_77":76,"_78":76,"_79":73,"_80":73,"_81":167,"_83":168,"_85":169,"_87":30},"tree",[157,158],"files",[157,158],"counts",{"_146":65,"_147":148,"_149":65,"_150":151,"_152":153},"contents",[130,131],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles",true,"grants","orgs","customDomains",{},{"_46":143,"_37":33,"_35":36,"_134":-5,"_44":45,"_43":14,"_136":125,"_41":144,"_13":65,"_38":26,"_39":40,"_29":30,"_141":-5,"_32":145},{"_46":132,"_37":133,"_35":60,"_134":-5,"_44":45,"_43":135,"_136":30,"_41":137,"_13":138,"_38":139,"_39":140,"_29":30,"_141":-5,"_32":142},"1373c79e-3103-11f1-bb4e-42dde27851f2","main.ts","parentId",7,"latest",["D",1775402470601],6,"http","/**\n * OpenAI-Compatible API Proxy — v2.0\n * ====================================\n * Non-streaming backend supporting:\n * • OpenAI models → via Val Town std/openai (no key needed)\n * • Claude models → via Anthropic SDK (requires ANTHROPIC_API_KEY env var)\n *\n * Both providers support tool_calls / function calling.\n *\n * Endpoints:\n * GET /api/v1/models — list all available models\n * POST /api/v1/chat/completions — non-streaming chat + tool_calls\n *\n * Auth : any Bearer token accepted (format-only check).\n * CORS : fully open.\n */\n\nimport { OpenAI } from \"https://esm.town/v/std/openai\";\nimport Anthropic from \"npm:@anthropic-ai/sdk\";\n\n// ─────────────────────────────────────────────────────────────────────────────\n// CORS helpers\n// ─────────────────────────────────────────────────────────────────────────────\n\nconst CORS_HEADERS: Record = {\n \"Access-Control-Allow-Origin\": \"*\",\n \"Access-Control-Allow-Methods\": \"GET, POST, OPTIONS\",\n \"Access-Control-Allow-Headers\": \"Content-Type, Authorization\",\n \"Access-Control-Max-Age\": \"86400\",\n};\n\nfunction corsJson(body: unknown, status = 200): Response {\n return new Response(JSON.stringify(body, null, 2), {\n status,\n headers: { ...CORS_HEADERS, \"Content-Type\": \"application/json\" },\n });\n}\n\nfunction corsError(message: string, status = 400): Response {\n return corsJson({ error: { message, type: \"invalid_request_error\", code: status } }, status);\n}\n\n// ─────────────────────────────────────────────────────────────────────────────\n// Model registry & routing\n// ─────────────────────────────────────────────────────────────────────────────\n\ntype Provider = \"openai\" | \"anthropic\";\n\ninterface ModelEntry {\n /** Canonical ID used internally */\n id: string;\n /** Human-readable name */\n name: string;\n /** Upstream model ID passed to the provider SDK */\n upstreamId: string;\n /** Which SDK to call */\n provider: Provider;\n}\n\nconst MODEL_REGISTRY: ModelEntry[] = [\n // ── OpenAI (via std/openai) ─────────────────────────────────────────────\n { id: \"gpt-4o\", name: \"GPT-4o\", upstreamId: \"gpt-4o\", provider: \"openai\" },\n { id: \"gpt-4o-mini\", name: \"GPT-4o Mini\", upstreamId: \"gpt-4o-mini\", provider: \"openai\" },\n { id: \"gpt-4.1\", name: \"GPT-4.1\", upstreamId: \"gpt-4.1\", provider: \"openai\" },\n { id: \"gpt-4.1-mini\", name: \"GPT-4.1 Mini\", upstreamId: \"gpt-4.1-mini\", provider: \"openai\" },\n { id: \"gpt-4.1-nano\", name: \"GPT-4.1 Nano\", upstreamId: \"gpt-4.1-nano\", provider: \"openai\" },\n { id: \"gpt-5-nano\", name: \"GPT-5 Nano\", upstreamId: \"gpt-5-nano\", provider: \"openai\" },\n\n // ── Anthropic Claude (via Anthropic SDK — needs ANTHROPIC_API_KEY) ───────\n { id: \"claude-opus-4\", name: \"Claude Opus 4\", upstreamId: \"claude-opus-4-5\", provider: \"anthropic\" },\n { id: \"claude-sonnet-4\", name: \"Claude Sonnet 4\", upstreamId: \"claude-sonnet-4-5\", provider: \"anthropic\" },\n { id: \"claude-haiku-3-5\", name: \"Claude Haiku 3.5\", upstreamId: \"claude-haiku-3-5-20241022\", provider: \"anthropic\" },\n { id: \"claude-3-5-sonnet\", name: \"Claude 3.5 Sonnet\", upstreamId: \"claude-3-5-sonnet-20241022\", provider: \"anthropic\" },\n { id: \"claude-3-5-haiku\", name: \"Claude 3.5 Haiku\", upstreamId: \"claude-3-5-haiku-20241022\", provider: \"anthropic\" },\n { id: \"claude-3-opus\", name: \"Claude 3 Opus\", upstreamId: \"claude-3-opus-20240229\", provider: \"anthropic\" },\n];\n\n/** Friendly alias → canonical id in MODEL_REGISTRY */\nconst ALIASES: Record = {\n // GPT aliases\n \"gpt-5\": \"gpt-4o\",\n \"gpt-5-turbo\": \"gpt-4o\",\n \"gpt-4-turbo\": \"gpt-4o\",\n \"gpt-4\": \"gpt-4o\",\n \"gpt-3.5-turbo\": \"gpt-4o-mini\",\n \"chatgpt\": \"gpt-4o\",\n \"chat-gpt\": \"gpt-4o\",\n\n // Claude 4.x friendly aliases\n \"claude-4-opus\": \"claude-opus-4\",\n \"claude-4-sonnet\": \"claude-sonnet-4\",\n \"claude-4.6-sonnet\": \"claude-sonnet-4\",\n \"claude-4.5-sonnet\": \"claude-sonnet-4\",\n \"claude-sonnet-4-5\": \"claude-sonnet-4\",\n \"anthropic-claude-sonnet-4-5\": \"claude-sonnet-4\",\n\n // Claude 3.x friendly aliases\n \"claude-3-sonnet\": \"claude-3-5-sonnet\",\n \"claude-3-haiku\": \"claude-haiku-3-5\",\n \"claude-instant\": \"claude-haiku-3-5\",\n\n // Gemini → best OpenAI fallback (Gemini not yet supported natively)\n \"gemini-2.5-pro\": \"gpt-4o\",\n \"gemini-1.5-pro\": \"gpt-4o\",\n \"gemini-pro\": \"gpt-4o\",\n \"google-gemini-2-5-pro\": \"gpt-4o\",\n};\n\nconst DEFAULT_ENTRY = MODEL_REGISTRY.find((m) => m.id === \"gpt-4o\")!;\n\nfunction resolveModel(requested: string): ModelEntry {\n // 1. Exact match in registry\n const exact = MODEL_REGISTRY.find((m) => m.id === requested);\n if (exact) return exact;\n\n // 2. Alias lookup\n const aliasTarget = ALIASES[requested];\n if (aliasTarget) {\n const entry = MODEL_REGISTRY.find((m) => m.id === aliasTarget);\n if (entry) {\n console.log(`[model-map] alias \"${requested}\" → \"${entry.id}\" (${entry.provider})`);\n return entry;\n }\n }\n\n // 3. Fallback\n console.log(`[model-map] unknown model \"${requested}\" → fallback \"${DEFAULT_ENTRY.id}\"`);\n return DEFAULT_ENTRY;\n}\n\n// ─────────────────────────────────────────────────────────────────────────────\n// Type helpers\n// ─────────────────────────────────────────────────────────────────────────────\n\ninterface OAIMessage {\n role: string;\n content: string | null;\n name?: string;\n tool_calls?: OAIToolCall[];\n tool_call_id?: string;\n}\n\ninterface OAIToolCall {\n id: string;\n type: \"function\";\n function: { name: string; arguments: string };\n}\n\ninterface OAITool {\n type: \"function\";\n function: {\n name: string;\n description?: string;\n parameters?: Record;\n };\n}\n\nfunction nowSec() { return Math.floor(Date.now() / 1000); }\nfunction randomId(p = \"chatcmpl\") { return `${p}-${crypto.randomUUID().replace(/-/g, \"\").slice(0, 24)}`; }\n\n// ─────────────────────────────────────────────────────────────────────────────\n// GET /api/v1/models\n// ─────────────────────────────────────────────────────────────────────────────\n\nfunction handleModels(): Response {\n const now = nowSec();\n const seen = new Set();\n const data: object[] = [];\n\n // Canonical entries\n for (const m of MODEL_REGISTRY) {\n seen.add(m.id);\n data.push({ id: m.id, object: \"model\", created: now, owned_by: m.provider });\n }\n // Alias entries\n for (const [alias, target] of Object.entries(ALIASES)) {\n if (seen.has(alias)) continue;\n seen.add(alias);\n const parent = MODEL_REGISTRY.find((m) => m.id === target);\n data.push({ id: alias, object: \"model\", created: now, owned_by: parent?.provider ?? \"openai\" });\n }\n\n return corsJson({ object: \"list\", data });\n}\n\n// ─────────────────────────────────────────────────────────────────────────────\n// OpenAI path (non-streaming + tool_calls)\n// ─────────────────────────────────────────────────────────────────────────────\n\nasync function callOpenAI(\n entry: ModelEntry,\n body: Record,\n): Promise {\n const openai = new OpenAI();\n\n // Build params — pass through all OpenAI-compatible fields\n // deno-lint-ignore no-explicit-any\n const params: any = {\n model: entry.upstreamId,\n messages: body.messages,\n temperature: body.temperature,\n max_tokens: body.max_tokens,\n top_p: body.top_p,\n stop: body.stop,\n stream: false,\n };\n\n // Forward tools / function calling if present\n if (body.tools) params.tools = body.tools;\n if (body.tool_choice) params.tool_choice = body.tool_choice;\n if (body.functions) params.functions = body.functions; // legacy\n if (body.function_call) params.function_call = body.function_call; // legacy\n\n // Strip undefined keys\n Object.keys(params).forEach((k) => params[k] === undefined && delete params[k]);\n\n const completion = await openai.chat.completions.create(params);\n\n // Return verbatim but override model to what caller requested\n return corsJson({ ...completion, model: body.model });\n}\n\n// ─────────────────────────────────────────────────────────────────────────────\n// Anthropic path (non-streaming + tool_calls)\n// ─────────────────────────────────────────────────────────────────────────────\n\n/**\n * Convert OpenAI messages array → Anthropic messages + optional system string.\n * Handles: text content, tool_calls (assistant), tool results (tool role).\n */\nfunction toAnthropicMessages(oaiMessages: OAIMessage[]): {\n system: string | undefined;\n messages: Anthropic.MessageParam[];\n} {\n let system: string | undefined;\n const messages: Anthropic.MessageParam[] = [];\n\n for (const msg of oaiMessages) {\n if (msg.role === \"system\") {\n // Anthropic takes system as a top-level param\n system = (system ? system + \"\\n\" : \"\") + (msg.content ?? \"\");\n continue;\n }\n\n if (msg.role === \"user\") {\n messages.push({ role: \"user\", content: msg.content ?? \"\" });\n continue;\n }\n\n if (msg.role === \"assistant\") {\n // May contain text and/or tool_use blocks\n const content: Anthropic.ContentBlock[] = [];\n\n if (msg.content) {\n content.push({ type: \"text\", text: msg.content });\n }\n\n if (msg.tool_calls && msg.tool_calls.length > 0) {\n for (const tc of msg.tool_calls) {\n let input: Record = {};\n try { input = JSON.parse(tc.function.arguments); } catch { /* keep empty */ }\n content.push({\n type: \"tool_use\",\n id: tc.id,\n name: tc.function.name,\n input,\n } as Anthropic.ToolUseBlock);\n }\n }\n\n messages.push({ role: \"assistant\", content: content.length ? content : \"\" });\n continue;\n }\n\n if (msg.role === \"tool\") {\n // OpenAI tool result → Anthropic tool_result block (must follow assistant)\n messages.push({\n role: \"user\",\n content: [\n {\n type: \"tool_result\",\n tool_use_id: msg.tool_call_id ?? \"\",\n content: msg.content ?? \"\",\n } as Anthropic.ToolResultBlockParam,\n ],\n });\n continue;\n }\n\n // function role (legacy) → treat as user\n if (msg.role === \"function\") {\n messages.push({\n role: \"user\",\n content: `Function ${msg.name ?? \"\"} returned: ${msg.content ?? \"\"}`,\n });\n }\n }\n\n return { system, messages };\n}\n\n/**\n * Convert OpenAI tools array → Anthropic tools array.\n */\nfunction toAnthropicTools(oaiTools: OAITool[]): Anthropic.Tool[] {\n return oaiTools.map((t) => ({\n name: t.function.name,\n description: t.function.description ?? \"\",\n input_schema: (t.function.parameters ?? { type: \"object\", properties: {} }) as Anthropic.Tool[\"input_schema\"],\n }));\n}\n\n/**\n * Convert Anthropic response → OpenAI response shape.\n * Handles both text replies and tool_use blocks.\n */\nfunction fromAnthropicResponse(\n response: Anthropic.Message,\n requestedModel: string,\n): Record {\n const completionId = randomId();\n const created = nowSec();\n\n // Separate text blocks from tool_use blocks\n const textBlocks = response.content.filter((b) => b.type === \"text\") as Anthropic.TextBlock[];\n const toolUseBlocks = response.content.filter((b) => b.type === \"tool_use\") as Anthropic.ToolUseBlock[];\n\n const textContent = textBlocks.map((b) => b.text).join(\"\") || null;\n\n // Build tool_calls array if there are tool_use blocks\n let tool_calls: OAIToolCall[] | undefined;\n if (toolUseBlocks.length > 0) {\n tool_calls = toolUseBlocks.map((b) => ({\n id: b.id,\n type: \"function\" as const,\n function: {\n name: b.name,\n arguments: JSON.stringify(b.input),\n },\n }));\n }\n\n // finish_reason mapping\n const finishMap: Record = {\n end_turn: \"stop\",\n max_tokens: \"length\",\n tool_use: \"tool_calls\",\n stop_sequence:\"stop\",\n };\n const finishReason = finishMap[response.stop_reason ?? \"end_turn\"] ?? \"stop\";\n\n return {\n id: completionId,\n object: \"chat.completion\",\n created,\n model: requestedModel,\n choices: [\n {\n index: 0,\n message: {\n role: \"assistant\",\n content: textContent,\n ...(tool_calls ? { tool_calls } : {}),\n },\n finish_reason: finishReason,\n },\n ],\n usage: {\n prompt_tokens: response.usage.input_tokens,\n completion_tokens: response.usage.output_tokens,\n total_tokens: response.usage.input_tokens + response.usage.output_tokens,\n },\n };\n}\n\nasync function callAnthropic(\n entry: ModelEntry,\n body: Record,\n): Promise {\n const apiKey = Deno.env.get(\"ANTHROPIC_API_KEY\");\n if (!apiKey) {\n return corsError(\n \"ANTHROPIC_API_KEY environment variable is not set. \" +\n \"Add it in the val's Environment variables settings to use Claude models.\",\n 503,\n );\n }\n\n const client = new Anthropic({ apiKey });\n\n const oaiMessages = body.messages as OAIMessage[];\n const { system, messages } = toAnthropicMessages(oaiMessages);\n\n // deno-lint-ignore no-explicit-any\n const params: any = {\n model: entry.upstreamId,\n messages,\n max_tokens: typeof body.max_tokens === \"number\" ? body.max_tokens : 4096,\n };\n\n if (system) params.system = system;\n if (body.temperature) params.temperature = body.temperature;\n if (body.top_p) params.top_p = body.top_p;\n if (body.stop) {\n params.stop_sequences = Array.isArray(body.stop) ? body.stop : [body.stop];\n }\n\n // Tool definitions\n const oaiTools = body.tools as OAITool[] | undefined;\n if (oaiTools && oaiTools.length > 0) {\n params.tools = toAnthropicTools(oaiTools);\n }\n\n // tool_choice mapping: \"auto\" | \"required\" | { type:\"function\", function:{name} }\n if (body.tool_choice) {\n const tc = body.tool_choice;\n if (tc === \"auto\") params.tool_choice = { type: \"auto\" };\n else if (tc === \"none\") { /* omit — Anthropic has no \"none\", just don't pass tools */ }\n else if (tc === \"required\") params.tool_choice = { type: \"any\" };\n else if (typeof tc === \"object\" && (tc as Record).type === \"function\") {\n const fnName = ((tc as Record).function as Record).name;\n params.tool_choice = { type: \"tool\", name: fnName };\n }\n }\n\n console.log(`[anthropic] calling \"${entry.upstreamId}\" tools=${params.tools?.length ?? 0}`);\n\n const response = await client.messages.create(params);\n const oaiResponse = fromAnthropicResponse(response, body.model as string);\n return corsJson(oaiResponse);\n}\n\n// ─────────────────────────────────────────────────────────────────────────────\n// Auth check\n// ─────────────────────────────────────────────────────────────────────────────\n\nfunction checkAuth(req: Request): string | null {\n const auth = req.headers.get(\"Authorization\") ?? \"\";\n if (!auth.startsWith(\"Bearer \")) {\n return \"Missing or malformed Authorization header. Expected: Bearer \";\n }\n if (auth.slice(7).trim().length === 0) {\n return \"Bearer token must not be empty.\";\n }\n return null;\n}\n\n// ─────────────────────────────────────────────────────────────────────────────\n// Main router\n// ─────────────────────────────────────────────────────────────────────────────\n\nexport default async function handler(req: Request): Promise {\n const url = new URL(req.url);\n const path = url.pathname;\n const method = req.method.toUpperCase();\n\n console.log(`[request] ${method} ${path}`);\n\n // CORS preflight\n if (method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: CORS_HEADERS });\n }\n\n // Health check / root\n if (path === \"/\" || path === \"\") {\n return corsJson({\n service: \"OpenAI-Compatible API Proxy\",\n version: \"2.0.0\",\n status: \"ok\",\n providers: [\"openai (built-in)\", \"anthropic (ANTHROPIC_API_KEY required)\"],\n endpoints: {\n models: \"GET /api/v1/models\",\n chat: \"POST /api/v1/chat/completions\",\n },\n });\n }\n\n // ── GET /api/v1/models ───────────────────────────────────────────────────\n if (path === \"/api/v1/models\" && method === \"GET\") {\n return handleModels();\n }\n\n // ── POST /api/v1/chat/completions ────────────────────────────────────────\n if (path === \"/api/v1/chat/completions\" && method === \"POST\") {\n // Auth\n const authErr = checkAuth(req);\n if (authErr) return corsError(authErr, 401);\n\n // Parse body\n let body: Record;\n try {\n body = await req.json();\n } catch {\n return corsError(\"Invalid JSON body.\", 400);\n }\n\n // Validate\n if (!body.messages || !Array.isArray(body.messages) || body.messages.length === 0) {\n return corsError(\"`messages` must be a non-empty array.\", 400);\n }\n\n // Reject streaming (non-streaming only in v2)\n if (body.stream === true) {\n return corsError(\n \"Streaming is disabled in this proxy. Please set stream=false (or omit stream).\",\n 400,\n );\n }\n\n const requestedModel = typeof body.model === \"string\" ? body.model : DEFAULT_ENTRY.id;\n const entry = resolveModel(requestedModel);\n\n console.log(\n `[chat] requested=\"${requestedModel}\" → entry=\"${entry.id}\" ` +\n `upstream=\"${entry.upstreamId}\" provider=\"${entry.provider}\" ` +\n `tools=${Array.isArray(body.tools) ? (body.tools as unknown[]).length : 0}`,\n );\n\n try {\n if (entry.provider === \"anthropic\") {\n return await callAnthropic(entry, body);\n } else {\n return await callOpenAI(entry, body);\n }\n } catch (err: unknown) {\n const msg = err instanceof Error ? err.message : String(err);\n console.error(`[error] ${msg}`);\n // Surface Anthropic API errors with their original message\n return corsError(`Upstream error (${entry.provider}): ${msg}`, 502);\n }\n }\n\n // 404\n return corsJson(\n { error: { message: `Route not found: ${method} ${path}`, type: \"invalid_request_error\", code: 404 } },\n 404,\n );\n}\n","label",[133],"2fa930de-3103-11f1-ba5c-42dde27851f2",["D",1775402517936],[33],"pulls","branches",1,"forks","envVars",{"_154":148,"_156":65},"connections",{"_154":65,"_155":65},"count","unfilledCount","emptyCount",{"_46":143,"_37":33,"_35":36,"_134":-5,"_44":45,"_43":14,"_136":125,"_41":163,"_13":65,"_38":26,"_39":40,"_29":30,"_141":-5,"_160":164,"_32":165},{"_46":132,"_37":133,"_35":60,"_134":-5,"_44":45,"_43":135,"_136":30,"_41":159,"_13":138,"_38":139,"_39":140,"_29":30,"_141":-5,"_160":161,"_32":162},["D",1775402470601],"children",[],[133],["D",1775402517936],[],[33],["D",1775401415177],{"_99":98,"_106":102},{"_99":98,"_100":-5,"_101":102,"_19":58,"_103":104,"_105":30},{"_13":88,"_37":89,"_46":45,"_90":47,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":48,"_97":98},"ctx",{"_81":-5,"_127":203,"_204":205,"_206":207,"_208":30,"_209":30,"_210":30,"_211":125,"_212":-5,"_213":-7,"_214":-7,"_215":65,"_216":30,"_217":30,"_218":-5,"_219":30,"_220":30,"_221":30,"_222":-5},"ENV",{"_174":175,"_176":177,"_178":89,"_179":180,"_181":182,"_183":184,"_185":186,"_187":188,"_189":190,"_191":192,"_193":194,"_195":196,"_197":198,"_199":200,"_201":202},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","71c1b20804b3232a7e513be260cfffc1932a9e97","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

Diff · v8