[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_190":191,"_192":193},"routes/_app.x.$handle.$name",{"_11":12,"_13":14,"_15":16,"_18":19,"_17":-7,"_144":-5,"_20":145,"_22":146,"_147":148,"_149":150,"_25":151,"_24":-5,"_152":-5,"_153":-5,"_154":55,"_155":156,"_157":46,"_158":-5,"_159":-5,"_160":161},"routes/_app.x.$handle.$name.branch.$branch.code.$",{"_11":12,"_13":14,"_15":16,"_17":-7,"_18":19,"_20":21,"_22":23,"_24":-7,"_25":26,"_27":-5,"_28":-5,"_29":-5,"_30":-5,"_31":-5},"handle","noppo0152","name","imagen-mcp-server","branchName","main","version","path","","project",{"_35":88,"_13":14,"_47":89,"_90":-5,"_91":55,"_92":93,"_94":93,"_95":96,"_97":-5,"_98":99,"_100":60,"_101":46,"_102":-5,"_103":-5,"_104":105,"_106":107,"_108":-5,"_109":-5,"_110":-5,"_111":-5,"_112":113,"_114":113,"_115":116,"_117":116,"_118":116,"_119":113,"_120":113,"_121":122,"_123":124,"_125":126,"_127":55},"tree",[75,76,77],"file","contents",[32,33,34],"upstream","pull","ahead","behind","envVars",{"_35":67,"_13":68,"_38":69,"_40":-5,"_41":42,"_43":70,"_45":46,"_47":71,"_17":72,"_50":24,"_52":73,"_54":55,"_56":-5,"_18":74},{"_35":58,"_13":59,"_38":60,"_40":-5,"_41":42,"_43":61,"_45":46,"_47":62,"_17":63,"_50":64,"_52":65,"_54":55,"_56":-5,"_18":66},{"_35":36,"_13":37,"_38":39,"_40":-5,"_41":42,"_43":44,"_45":46,"_47":48,"_17":49,"_50":51,"_52":53,"_54":55,"_56":-5,"_18":57},"id","527986c6-6239-4fb9-840c-fcc2a401fdae","tools.ts","fileId","4c799d92-89b0-11f1-90a6-1607ee4eb77e","parentId","branchId","ebf27ee7-a197-4e55-9ee2-32abfef5065f","branchVersion",8,"latest",true,"createdAt",["D",1785152656184],3,"fileType","script","content","// Tool definitions and API integrations for the Gemini MCP Server\n\nconst IMAGEN_MODELS = {\n \"imagen-4.0-generate-001\": \"Imagen 4 (highest quality)\",\n \"imagen-4.0-fast-generate-001\": \"Imagen 4 Fast (lower latency)\",\n \"imagen-3.0-generate-002\": \"Imagen 3\",\n} as const;\n\nconst VEO_MODELS = {\n \"veo-3.1-generate-preview\": \"Veo 3.1 (latest, native audio)\",\n} as const;\n\nconst GEMINI_VISION_MODELS = {\n \"gemini-2.5-flash\": \"Gemini 2.5 Flash (recommended)\",\n \"gemini-2.0-flash\": \"Gemini 2.0 Flash\",\n \"gemini-1.5-pro\": \"Gemini 1.5 Pro (may be deprecated)\",\n} as const;\n\nfunction getApiKey(): string {\n const apiKey = Deno.env.get(\"GEMINI_API_KEY\");\n if (!apiKey) {\n throw new Error(\"GEMINI_API_KEY environment variable is not set.\");\n }\n return apiKey;\n}\n\n// ---------------------------------------------------------------------------\n// 1. generate_image — Imagen text-to-image (existing)\n// ---------------------------------------------------------------------------\n\nasync function generateImage(params: {\n prompt: string;\n model?: string;\n sampleCount?: number;\n aspectRatio?: string;\n personGeneration?: string;\n imageSize?: string;\n seed?: number;\n}) {\n const apiKey = getApiKey();\n const model = params.model ?? \"imagen-4.0-generate-001\";\n const url =\n `https://generativelanguage.googleapis.com/v1beta/models/${model}:predict`;\n\n const parameters: Record = {\n sampleCount: params.sampleCount ?? 1,\n aspectRatio: params.aspectRatio ?? \"1:1\",\n personGeneration: params.personGeneration ?? \"allow_adult\",\n };\n if (params.imageSize) parameters.imageSize = params.imageSize;\n if (params.seed !== undefined) parameters.seed = params.seed;\n\n const resp = await fetch(url, {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/json\", \"x-goog-api-key\": apiKey },\n body: JSON.stringify({\n instances: [{ prompt: params.prompt }],\n parameters,\n }),\n });\n\n if (!resp.ok) {\n const errorText = await resp.text();\n let message = `Imagen API error (${resp.status})`;\n try {\n message = JSON.parse(errorText).error?.message ?? message;\n } catch {\n message = `${message}: ${errorText}`;\n }\n throw new Error(message);\n }\n\n const data = await resp.json();\n return data.predictions as Array<{\n bytesBase64Encoded: string;\n mimeType: string;\n }>;\n}\n\n// ---------------------------------------------------------------------------\n// 2. edit_image — Imagen 3 image-to-image editing\n// ---------------------------------------------------------------------------\n\nasync function editImage(params: {\n prompt: string;\n image: string; // base64-encoded source image\n mimeType: string;\n model?: string;\n sampleCount?: number;\n aspectRatio?: string;\n personGeneration?: string;\n}) {\n const apiKey = getApiKey();\n const model = params.model ?? \"imagen-3.0-generate-002\";\n const url =\n `https://generativelanguage.googleapis.com/v1beta/models/${model}:predict`;\n\n const parameters: Record = {\n sampleCount: params.sampleCount ?? 1,\n aspectRatio: params.aspectRatio ?? \"1:1\",\n personGeneration: params.personGeneration ?? \"allow_adult\",\n };\n\n const resp = await fetch(url, {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/json\", \"x-goog-api-key\": apiKey },\n body: JSON.stringify({\n instances: [\n {\n prompt: params.prompt,\n image: {\n bytesBase64Encoded: params.image,\n mimeType: params.mimeType,\n },\n },\n ],\n parameters,\n }),\n });\n\n if (!resp.ok) {\n const errorText = await resp.text();\n let message = `Imagen edit API error (${resp.status})`;\n try {\n message = JSON.parse(errorText).error?.message ?? message;\n } catch {\n message = `${message}: ${errorText}`;\n }\n throw new Error(message);\n }\n\n const data = await resp.json();\n return data.predictions as Array<{\n bytesBase64Encoded: string;\n mimeType: string;\n }>;\n}\n\n// ---------------------------------------------------------------------------\n// 3. generate_video — Veo 3.1 text-to-video (long-running operation)\n// ---------------------------------------------------------------------------\n\nconst VEO_BASE = \"https://generativelanguage.googleapis.com/v1beta\";\nconst VEO_MAX_POLLS = 60; // 60 × 10s = 10 min max wait\n\nasync function generateVideo(params: {\n prompt: string;\n model?: string;\n aspectRatio?: string;\n resolution?: string;\n}): Promise<{\n videoBase64: string;\n mimeType: string;\n operationName: string;\n}> {\n const apiKey = getApiKey();\n const model = params.model ?? \"veo-3.1-generate-preview\";\n\n // Step 1 — submit the long-running operation\n const submitResp = await fetch(\n `${VEO_BASE}/models/${model}:predictLongRunning`,\n {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"x-goog-api-key\": apiKey,\n },\n body: JSON.stringify({\n instances: [{ prompt: params.prompt }],\n parameters: {\n aspectRatio: params.aspectRatio ?? \"16:9\",\n resolution: params.resolution ?? \"720p\",\n },\n }),\n },\n );\n\n if (!submitResp.ok) {\n const errorText = await submitResp.text();\n let message = `Veo API error (${submitResp.status})`;\n try {\n message = JSON.parse(errorText).error?.message ?? message;\n } catch {\n message = `${message}: ${errorText}`;\n }\n throw new Error(message);\n }\n\n const submitData = await submitResp.json();\n const operationName = submitData.name as string;\n if (!operationName) {\n throw new Error(\"Veo API did not return an operation name.\");\n }\n\n // Step 2 — poll until the operation completes\n for (let i = 0; i < VEO_MAX_POLLS; i++) {\n await new Promise((resolve) => setTimeout(resolve, 10_000));\n\n const pollResp = await fetch(`${VEO_BASE}/${operationName}`, {\n headers: { \"x-goog-api-key\": apiKey },\n });\n\n if (!pollResp.ok) {\n const errorText = await pollResp.text();\n throw new Error(`Veo polling error (${pollResp.status}): ${errorText}`);\n }\n\n const pollData = await pollResp.json();\n\n if (pollData.error) {\n throw new Error(\n `Veo operation failed: ${pollData.error.message ?? JSON.stringify(pollData.error)}`,\n );\n }\n\n if (pollData.done) {\n const samples = pollData.response?.generateVideoResponse\n ?.generatedSamples as Array<{ video: { uri: string } }> | undefined;\n if (!samples || samples.length === 0) {\n throw new Error(\"Veo operation completed but no video was returned.\");\n }\n\n // Step 3 — download the generated video\n const videoUri = samples[0].video.uri;\n const downloadResp = await fetch(videoUri, {\n headers: { \"x-goog-api-key\": apiKey },\n });\n\n if (!downloadResp.ok) {\n throw new Error(\n `Failed to download video (${downloadResp.status})`,\n );\n }\n\n const videoBuffer = new Uint8Array(\n await downloadResp.arrayBuffer(),\n );\n const videoBase64 = btoa(\n String.fromCharCode(...videoBuffer),\n );\n\n return {\n videoBase64,\n mimeType: \"video/mp4\",\n operationName,\n };\n }\n }\n\n throw new Error(\n `Veo video generation timed out after ${VEO_MAX_POLLS * 10} seconds. ` +\n `Operation: ${operationName}`,\n );\n}\n\n// ---------------------------------------------------------------------------\n// 4. analyze_image — Gemini vision (image understanding + prompt engineering)\n// ---------------------------------------------------------------------------\n\nasync function analyzeImage(params: {\n image: string; // base64-encoded image\n mimeType: string;\n prompt?: string;\n model?: string;\n}): Promise {\n const apiKey = getApiKey();\n const model = params.model ?? \"gemini-2.5-flash\";\n const url =\n `https://generativelanguage.googleapis.com/v1beta/models/${model}:generateContent`;\n\n const textPrompt = params.prompt ??\n \"Analyze this image in detail. Describe the subject, composition, colors, style, mood, and any notable elements. Then suggest an enhanced prompt that could be used to generate a similar image with an AI image generator.\";\n\n const resp = await fetch(url, {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/json\", \"x-goog-api-key\": apiKey },\n body: JSON.stringify({\n contents: [\n {\n parts: [\n { text: textPrompt },\n {\n inline_data: {\n mime_type: params.mimeType,\n data: params.image,\n },\n },\n ],\n },\n ],\n generationConfig: {\n temperature: 0.4,\n maxOutputTokens: 2048,\n },\n }),\n });\n\n if (!resp.ok) {\n const errorText = await resp.text();\n let message = `Gemini API error (${resp.status})`;\n try {\n message = JSON.parse(errorText).error?.message ?? message;\n } catch {\n message = `${message}: ${errorText}`;\n }\n throw new Error(message);\n }\n\n const data = await resp.json();\n const text = data.candidates?.[0]?.content?.parts?.[0]?.text;\n if (!text) {\n throw new Error(\"Gemini returned no text output.\");\n }\n return text as string;\n}\n\n// ---------------------------------------------------------------------------\n// Tool definitions (MCP schema)\n// ---------------------------------------------------------------------------\n\nexport const TOOLS = [\n {\n name: \"generate_image\",\n description:\n \"Generate an image from a text prompt using Google Imagen. Returns one or more base64-encoded images. \" +\n \"Note: Imagen models are deprecated and will be shut down on August 17, 2026; \" +\n \"consider migrating to Nano Banana (gemini-2.5-flash-image) for future use.\",\n inputSchema: {\n type: \"object\",\n properties: {\n prompt: {\n type: \"string\",\n description:\n \"A text description of the image to generate. English only. Max 480 tokens.\",\n },\n model: {\n type: \"string\",\n enum: Object.keys(IMAGEN_MODELS),\n description:\n `The Imagen model to use. Options: ${\n Object.entries(IMAGEN_MODELS).map(([k, v]) => `${k} (${v})`).join(\n \", \",\n )\n }. Defaults to imagen-4.0-generate-001.`,\n },\n sampleCount: {\n type: \"number\",\n description: \"Number of images to generate (1-4). Defaults to 1.\",\n minimum: 1,\n maximum: 4,\n },\n aspectRatio: {\n type: \"string\",\n enum: [\"1:1\", \"16:9\", \"9:16\", \"4:3\", \"3:4\"],\n description: \"Aspect ratio of the generated image. Defaults to 1:1.\",\n },\n personGeneration: {\n type: \"string\",\n enum: [\"dont_allow\", \"allow_adult\", \"allow_all\"],\n description: \"Controls generation of people. Defaults to allow_adult.\",\n },\n imageSize: {\n type: \"string\",\n enum: [\"1K\", \"2K\"],\n description:\n \"Output image resolution. Only supported for imagen-4.0-generate-001. Defaults to 1K.\",\n },\n seed: {\n type: \"number\",\n description:\n \"Random seed for reproducible generation. If not set, a random seed is used.\",\n minimum: 0,\n },\n },\n required: [\"prompt\"],\n },\n },\n {\n name: \"edit_image\",\n description:\n \"Edit or modify an existing image using a text prompt with Google Imagen 3. \" +\n \"Provide a base64-encoded source image and a description of the desired changes. \" +\n \"Returns one or more base64-encoded edited images.\",\n inputSchema: {\n type: \"object\",\n properties: {\n prompt: {\n type: \"string\",\n description:\n \"A text description of the edits to apply to the source image.\",\n },\n image: {\n type: \"string\",\n description:\n \"Base64-encoded source image data (without data URI prefix).\",\n },\n mimeType: {\n type: \"string\",\n enum: [\"image/png\", \"image/jpeg\", \"image/webp\"],\n description: \"MIME type of the source image.\",\n },\n model: {\n type: \"string\",\n enum: Object.keys(IMAGEN_MODELS),\n description:\n `The Imagen model to use. Options: ${\n Object.entries(IMAGEN_MODELS).map(([k, v]) => `${k} (${v})`).join(\n \", \",\n )\n }. Defaults to imagen-3.0-generate-002.`,\n },\n sampleCount: {\n type: \"number\",\n description: \"Number of edited images to generate (1-4). Defaults to 1.\",\n minimum: 1,\n maximum: 4,\n },\n aspectRatio: {\n type: \"string\",\n enum: [\"1:1\", \"16:9\", \"9:16\", \"4:3\", \"3:4\"],\n description: \"Aspect ratio of the output image. Defaults to 1:1.\",\n },\n personGeneration: {\n type: \"string\",\n enum: [\"dont_allow\", \"allow_adult\", \"allow_all\"],\n description: \"Controls generation of people. Defaults to allow_adult.\",\n },\n },\n required: [\"prompt\", \"image\", \"mimeType\"],\n },\n },\n {\n name: \"generate_video\",\n description:\n \"Generate a short video (8 seconds, with native audio) from a text prompt using Google Veo 3.1. \" +\n \"This is a long-running operation that may take 1-5 minutes. \" +\n \"Returns the video as base64-encoded MP4 data.\",\n inputSchema: {\n type: \"object\",\n properties: {\n prompt: {\n type: \"string\",\n description:\n \"A detailed text description of the video to generate. Include scene description, camera angles, dialogue, and sound effects.\",\n },\n model: {\n type: \"string\",\n enum: Object.keys(VEO_MODELS),\n description:\n `The Veo model to use. Options: ${\n Object.entries(VEO_MODELS).map(([k, v]) => `${k} (${v})`).join(\", \")\n }. Defaults to veo-3.1-generate-preview.`,\n },\n aspectRatio: {\n type: \"string\",\n enum: [\"16:9\", \"9:16\"],\n description:\n \"Aspect ratio: 16:9 (landscape) or 9:16 (portrait). Defaults to 16:9.\",\n },\n resolution: {\n type: \"string\",\n enum: [\"720p\", \"1080p\", \"4k\"],\n description: \"Output video resolution. Defaults to 720p.\",\n },\n },\n required: [\"prompt\"],\n },\n },\n {\n name: \"analyze_image\",\n description:\n \"Analyze an image using Google Gemini vision models. Can describe the image, answer questions about it, \" +\n \"perform object detection, or generate enhanced prompts for image generation. \" +\n \"Returns a detailed text analysis.\",\n inputSchema: {\n type: \"object\",\n properties: {\n image: {\n type: \"string\",\n description:\n \"Base64-encoded image data (without data URI prefix).\",\n },\n mimeType: {\n type: \"string\",\n enum: [\"image/png\", \"image/jpeg\", \"image/webp\", \"image/heic\", \"image/heif\"],\n description: \"MIME type of the image.\",\n },\n prompt: {\n type: \"string\",\n description:\n \"Question or instruction about the image. If omitted, the model will describe the image and suggest an enhanced generation prompt.\",\n },\n model: {\n type: \"string\",\n enum: Object.keys(GEMINI_VISION_MODELS),\n description:\n `The Gemini vision model to use. Options: ${\n Object.entries(GEMINI_VISION_MODELS).map(([k, v]) => `${k} (${v})`)\n .join(\", \")\n }. Defaults to gemini-2.5-flash.`,\n },\n },\n required: [\"image\", \"mimeType\"],\n },\n },\n];\n\n// ---------------------------------------------------------------------------\n// Tool call dispatcher\n// ---------------------------------------------------------------------------\n\nexport async function handleToolCall(\n name: string,\n args: Record,\n): Promise<{ content: Array> }> {\n switch (name) {\n case \"generate_image\": {\n const predictions = await generateImage({\n prompt: args.prompt as string,\n model: args.model as string | undefined,\n sampleCount: args.sampleCount as number | undefined,\n aspectRatio: args.aspectRatio as string | undefined,\n personGeneration: args.personGeneration as string | undefined,\n imageSize: args.imageSize as string | undefined,\n seed: args.seed as number | undefined,\n });\n\n const content = predictions.map((p) => ({\n type: \"image\" as const,\n data: p.bytesBase64Encoded,\n mimeType: p.mimeType,\n }));\n\n return { content };\n }\n\n case \"edit_image\": {\n const predictions = await editImage({\n prompt: args.prompt as string,\n image: args.image as string,\n mimeType: args.mimeType as string,\n model: args.model as string | undefined,\n sampleCount: args.sampleCount as number | undefined,\n aspectRatio: args.aspectRatio as string | undefined,\n personGeneration: args.personGeneration as string | undefined,\n });\n\n const content = predictions.map((p) => ({\n type: \"image\" as const,\n data: p.bytesBase64Encoded,\n mimeType: p.mimeType,\n }));\n\n return { content };\n }\n\n case \"generate_video\": {\n const result = await generateVideo({\n prompt: args.prompt as string,\n model: args.model as string | undefined,\n aspectRatio: args.aspectRatio as string | undefined,\n resolution: args.resolution as string | undefined,\n });\n\n return {\n content: [\n {\n type: \"text\" as const,\n text:\n `Video generated successfully.\\n` +\n `Format: ${result.mimeType}\\n` +\n `Operation: ${result.operationName}`,\n },\n {\n type: \"image\" as const,\n data: result.videoBase64,\n mimeType: result.mimeType,\n },\n ],\n };\n }\n\n case \"analyze_image\": {\n const text = await analyzeImage({\n image: args.image as string,\n mimeType: args.mimeType as string,\n prompt: args.prompt as string | undefined,\n model: args.model as string | undefined,\n });\n\n return {\n content: [\n {\n type: \"text\" as const,\n text,\n },\n ],\n };\n }\n\n default:\n throw new Error(`Unknown tool: ${name}`);\n }\n}\n","deleted",false,"label",[37],"e450bf72-2907-4cfd-aadb-3763c4d2c835","main.ts","fead52e6-89ac-11f1-be12-1607ee4eb77e",5,["D",1785152589299],2,"http","import { TOOLS, handleToolCall } from \"./tools.ts\";\n\nconst SERVER_INFO = {\n name: \"gemini-mcp-server\",\n version: \"2.0.0\",\n};\n\nconst PROTOCOL_VERSION = \"2024-11-05\";\n\nfunction jsonRpcResponse(id: unknown, result: unknown) {\n return { jsonrpc: \"2.0\", id, result };\n}\n\nfunction jsonRpcError(id: unknown, code: number, message: string) {\n return { jsonrpc: \"2.0\", id, error: { code, message } };\n}\n\nexport default async function handler(req: Request): Promise {\n if (req.method === \"GET\") {\n // SSE endpoint — minimal acknowledgment for streamable HTTP transport\n const body = new ReadableStream({\n start(controller) {\n controller.close();\n },\n });\n return new Response(body, {\n headers: {\n \"Content-Type\": \"text/event-stream\",\n \"Cache-Control\": \"no-cache\",\n },\n });\n }\n\n if (req.method !== \"POST\") {\n return new Response(\"Method Not Allowed\", { status: 405 });\n }\n\n let message: Record;\n try {\n message = await req.json();\n } catch {\n return new Response(\n JSON.stringify(\n jsonRpcError(null, -32700, \"Parse error: invalid JSON\"),\n ),\n { status: 400, headers: { \"Content-Type\": \"application/json\" } },\n );\n }\n\n const id = message.id ?? null;\n const method = message.method as string;\n const params = (message.params ?? {}) as Record;\n\n // Notifications (no id) — just acknowledge\n if (id === null) {\n return new Response(null, { status: 202 });\n }\n\n try {\n switch (method) {\n case \"initialize\": {\n const sessionId = crypto.randomUUID();\n const result = jsonRpcResponse(id, {\n protocolVersion: PROTOCOL_VERSION,\n capabilities: { tools: {} },\n serverInfo: SERVER_INFO,\n });\n return new Response(JSON.stringify(result), {\n status: 200,\n headers: {\n \"Content-Type\": \"application/json\",\n \"Mcp-Session-Id\": sessionId,\n },\n });\n }\n\n case \"tools/list\": {\n const result = jsonRpcResponse(id, { tools: TOOLS });\n return new Response(JSON.stringify(result), {\n status: 200,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n case \"tools/call\": {\n const toolName = params.name as string;\n const toolArgs = (params.arguments ?? {}) as Record<\n string,\n unknown\n >;\n try {\n const toolResult = await handleToolCall(toolName, toolArgs);\n const result = jsonRpcResponse(id, toolResult);\n return new Response(JSON.stringify(result), {\n status: 200,\n headers: { \"Content-Type\": \"application/json\" },\n });\n } catch (err) {\n const result = jsonRpcResponse(id, {\n content: [\n {\n type: \"text\",\n text: `Error: ${err instanceof Error ? err.message : String(err)}`,\n },\n ],\n isError: true,\n });\n return new Response(JSON.stringify(result), {\n status: 200,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n }\n\n case \"ping\": {\n const result = jsonRpcResponse(id, {});\n return new Response(JSON.stringify(result), {\n status: 200,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n default: {\n const result = jsonRpcError(id, -32601, `Method not found: ${method}`);\n return new Response(JSON.stringify(result), {\n status: 200,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n }\n } catch (err) {\n const result = jsonRpcError(\n id,\n -32603,\n err instanceof Error ? err.message : \"Internal error\",\n );\n return new Response(JSON.stringify(result), {\n status: 500,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n}\n",[59],"eabacc20-22f3-4e5c-a9cf-3bd4a51b6452","README.md","7b2535fa-89ad-11f1-b4dc-1607ee4eb77e",9,["D",1785152675081],1,"# Gemini MCP Server\n\nAn [MCP (Model Context Protocol)](https://modelcontextprotocol.io/) server exposing Google's Gemini API — image generation (Imagen), image editing (Imagen 3), video generation (Veo 3.1), and image analysis (Gemini vision) — as tools for Claude and other MCP clients.\n\nDeployed on Val Town via the MCP Streamable HTTP transport.\n\n## Quick Start\n\n### 1. Set your API key\n\n👉 Add GEMINI_API_KEY here: https://www.val.town/x/noppo0152/imagen-mcp-server/environment-variables?key=GEMINI_API_KEY\n\nGet a key from [Google AI Studio](https://aistudio.google.com/apikey).\n\n### 2. Configure your MCP client\n\n```json\n{\n \"mcpServers\": {\n \"gemini\": {\n \"url\": \"https://noppo0152--fead52e689ac11f1be121607ee4eb77e.web.val.run\",\n \"transport\": \"http\"\n }\n }\n}\n```\n\n## Available Tools\n\n### `generate_image`\n\nText-to-image generation with Google Imagen.\n\n| Parameter | Type | Required | Description |\n|---|---|---|---|\n| `prompt` | string | ✅ | Image description (English, max 480 tokens) |\n| `model` | string | ❌ | `imagen-4.0-generate-001` (default), `imagen-4.0-fast-generate-001`, `imagen-3.0-generate-002` |\n| `sampleCount` | number | ❌ | 1–4 images. Default: 1 |\n| `aspectRatio` | string | ❌ | `1:1`, `16:9`, `9:16`, `4:3`, `3:4` |\n| `personGeneration` | string | ❌ | `dont_allow`, `allow_adult`, `allow_all` |\n| `imageSize` | string | ❌ | `1K` or `2K` (only Imagen 4) |\n| `seed` | number | ❌ | Reproducible generation seed |\n\n### `edit_image`\n\nImage-to-image editing with Google Imagen 3.\n\n| Parameter | Type | Required | Description |\n|---|---|---|---|\n| `prompt` | string | ✅ | Edit description |\n| `image` | string | ✅ | Base64-encoded source image (no data URI prefix) |\n| `mimeType` | string | ✅ | `image/png`, `image/jpeg`, `image/webp` |\n| `model` | string | ❌ | Defaults to `imagen-3.0-generate-002` |\n| `sampleCount` | number | ❌ | 1–4. Default: 1 |\n| `aspectRatio` | string | ❌ | `1:1`, `16:9`, `9:16`, `4:3`, `3:4` |\n| `personGeneration` | string | ❌ | `dont_allow`, `allow_adult`, `allow_all` |\n\n### `generate_video`\n\nText-to-video generation with Google Veo 3.1 (8s clips with native audio).\n\n| Parameter | Type | Required | Description |\n|---|---|---|---|\n| `prompt` | string | ✅ | Detailed scene description with camera, dialogue, sound |\n| `model` | string | ❌ | `veo-3.1-generate-preview` (default) |\n| `aspectRatio` | string | ❌ | `16:9` (landscape, default) or `9:16` (portrait) |\n| `resolution` | string | ❌ | `720p` (default), `1080p`, `4k` |\n\n> ⏱️ This is a long-running operation (1–5 min). The server polls until completion and returns base64-encoded MP4 data.\n\n### `analyze_image`\n\nImage analysis and prompt engineering with Gemini vision models.\n\n| Parameter | Type | Required | Description |\n|---|---|---|---|\n| `image` | string | ✅ | Base64-encoded image (no data URI prefix) |\n| `mimeType` | string | ✅ | `image/png`, `image/jpeg`, `image/webp`, `image/heic`, `image/heif` |\n| `prompt` | string | ❌ | Question or instruction. Default: describe image + suggest enhanced generation prompt |\n| `model` | string | ❌ | `gemini-2.5-flash` (default), `gemini-2.0-flash`, `gemini-1.5-pro` |\n\n## MCP Protocol\n\nImplements the MCP Streamable HTTP transport (JSON-RPC 2.0):\n\n- **POST** — `initialize`, `tools/list`, `tools/call`, `ping`\n- **GET** — SSE endpoint for server-to-client streaming\n\n## Architecture\n\n```\nmain.ts — MCP protocol handler (JSON-RPC, HTTP transport)\ntools.ts — Tool definitions + Google API integrations\n```\n\n## ⚠️ Deprecation Notice\n\nImagen models are **deprecated** and will shut down on **August 17, 2026**. Google recommends migrating to [Nano Banana](https://ai.google.dev/gemini-api/docs/image-generation) (`gemini-2.5-flash-image`) for future image generation needs.\n",[68],{"_35":67,"_13":68,"_38":69,"_40":-5,"_41":42,"_43":70,"_45":46,"_47":85,"_17":72,"_50":24,"_52":73,"_54":55,"_56":-5,"_79":86,"_18":87},{"_35":58,"_13":59,"_38":60,"_40":-5,"_41":42,"_43":61,"_45":46,"_47":82,"_17":63,"_50":64,"_52":65,"_54":55,"_56":-5,"_79":83,"_18":84},{"_35":36,"_13":37,"_38":39,"_40":-5,"_41":42,"_43":44,"_45":46,"_47":78,"_17":49,"_50":51,"_52":53,"_54":55,"_56":-5,"_79":80,"_18":81},["D",1785152656184],"children",[],[37],["D",1785152589299],[],[59],["D",1785152675081],[],[68],"10b0660d-c4d7-4579-bd91-d317f84c08ba",["D",1785151100183],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","MCP server for Google Imagen image generation","imageUrl","userId","e16eef30-ad59-4ff6-8f72-02d04b373a30","pinnedFileId","pinPreview","pinMigrated","screenshotVersion","likeCount",0,"orgId","019fa349-663b-741c-a208-22ba1fe5a062","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_11":12,"_143":139},"org",{"_11":12,"_136":137,"_138":139,"_98":99,"_140":141,"_142":55},"branch",{"_17":70,"_13":16,"_35":42,"_128":88,"_129":-5,"_130":-5,"_131":-5,"_132":-5,"_133":-5,"_134":14,"_135":12},"isOwned","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","displayName","原田淳一","avatar","https://lh3.googleusercontent.com/a/ACg8ocKynJaldQAlV9JQ9eZTgg1BovEQEb7X7sDs2nG64t0McgpqQQ=s96-c","tier","free","canRestrictVals","profileImageUrl","adoptOrgId",{"_35":88,"_13":14,"_47":186,"_90":-5,"_91":55,"_92":93,"_94":93,"_95":96,"_97":-5,"_98":99,"_100":60,"_101":46,"_102":-5,"_103":-5,"_104":105,"_106":107,"_108":-5,"_109":-5,"_110":-5,"_111":-5,"_112":113,"_114":113,"_115":116,"_117":116,"_118":116,"_119":113,"_120":113,"_121":187,"_123":188,"_125":189,"_127":55},[177,178,179],"files",[177,178,179],"counts",{"_168":105,"_169":72,"_170":105,"_31":171,"_172":173},[162,163,164],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_35":67,"_13":68,"_38":69,"_40":-5,"_41":42,"_43":70,"_45":46,"_47":71,"_17":72,"_50":24,"_52":73,"_54":55,"_56":-5,"_18":167},{"_35":58,"_13":59,"_38":60,"_40":-5,"_41":42,"_43":61,"_45":46,"_47":62,"_17":63,"_50":64,"_52":65,"_54":55,"_56":-5,"_18":166},{"_35":36,"_13":37,"_38":39,"_40":-5,"_41":42,"_43":44,"_45":46,"_47":48,"_17":49,"_50":51,"_52":53,"_54":55,"_56":-5,"_18":165},[37],[59],[68],"pulls","branches","forks",{"_174":72,"_176":105},"connections",{"_174":105,"_175":105},"count","unfilledCount","emptyCount",{"_35":67,"_13":68,"_38":69,"_40":-5,"_41":42,"_43":70,"_45":46,"_47":85,"_17":72,"_50":24,"_52":73,"_54":55,"_56":-5,"_79":184,"_18":185},{"_35":58,"_13":59,"_38":60,"_40":-5,"_41":42,"_43":61,"_45":46,"_47":82,"_17":63,"_50":64,"_52":65,"_54":55,"_56":-5,"_79":182,"_18":183},{"_35":36,"_13":37,"_38":39,"_40":-5,"_41":42,"_43":44,"_45":46,"_47":78,"_17":49,"_50":51,"_52":53,"_54":55,"_56":-5,"_79":180,"_18":181},[],[37],[],[59],[],[68],["D",1785151100183],{"_11":12,"_143":139},{"_11":12,"_136":137,"_138":139,"_98":99,"_140":141,"_142":55},{"_17":70,"_13":16,"_35":42,"_128":88,"_129":-5,"_130":-5,"_131":-5,"_132":-5,"_133":-5,"_134":14,"_135":12},"ctx",{"_121":-5,"_159":223,"_224":225,"_226":227,"_228":55,"_229":55,"_230":55,"_231":46,"_232":-5,"_233":-7,"_234":-7,"_235":105,"_236":55,"_237":55,"_238":-5,"_239":55,"_240":55,"_241":55,"_242":-5},"ENV",{"_194":195,"_196":197,"_198":16,"_199":200,"_201":202,"_203":204,"_205":206,"_207":208,"_209":210,"_211":212,"_213":214,"_215":216,"_217":218,"_219":220,"_221":222},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","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"]