@chrischall/gemini-mcp 0.1.0 → 0.3.0

package/dist/bundle.js

@@ -2980,7 +2980,7 @@ var require_compile = __commonJS({
       const schOrFunc = root.refs[ref];
       if (schOrFunc)
         return schOrFunc;
-      let _sch = resolve.call(this, root, ref);
+      let _sch = resolve2.call(this, root, ref);
       if (_sch === void 0) {
         const schema = (_a3 = root.localRefs) === null || _a3 === void 0 ? void 0 : _a3[ref];
         const { schemaId } = this.opts;
@@ -3007,7 +3007,7 @@ var require_compile = __commonJS({
     function sameSchemaEnv(s1, s2) {
       return s1.schema === s2.schema && s1.root === s2.root && s1.baseId === s2.baseId;
     }
-    function resolve(root, ref) {
+    function resolve2(root, ref) {
       let sch;
       while (typeof (sch = this.refs[ref]) == "string")
         ref = sch;
@@ -3638,7 +3638,7 @@ var require_fast_uri = __commonJS({
       }
       return uri;
     }
-    function resolve(baseURI, relativeURI, options) {
+    function resolve2(baseURI, relativeURI, options) {
       const schemelessOptions = options ? Object.assign({ scheme: "null" }, options) : { scheme: "null" };
       const resolved = resolveComponent(parse3(baseURI, schemelessOptions), parse3(relativeURI, schemelessOptions), schemelessOptions, true);
       schemelessOptions.skipEscape = true;
@@ -3896,7 +3896,7 @@ var require_fast_uri = __commonJS({
     var fastUri = {
       SCHEMES,
       normalize,
-      resolve,
+      resolve: resolve2,
       resolveComponent,
       equal,
       serialize,
@@ -28827,7 +28827,7 @@ var Protocol = class {
           return;
         }
         const pollInterval = task2.pollInterval ?? this._options?.defaultTaskPollInterval ?? 1e3;
-        await new Promise((resolve) => setTimeout(resolve, pollInterval));
+        await new Promise((resolve2) => setTimeout(resolve2, pollInterval));
         options?.signal?.throwIfAborted();
       }
     } catch (error51) {
@@ -28844,7 +28844,7 @@ var Protocol = class {
    */
   request(request, resultSchema, options) {
     const { relatedRequestId, resumptionToken, onresumptiontoken, task, relatedTask } = options ?? {};
-    return new Promise((resolve, reject) => {
+    return new Promise((resolve2, reject) => {
       const earlyReject = (error51) => {
         reject(error51);
       };
@@ -28922,7 +28922,7 @@ var Protocol = class {
           if (!parseResult.success) {
             reject(parseResult.error);
           } else {
-            resolve(parseResult.data);
+            resolve2(parseResult.data);
           }
         } catch (error51) {
           reject(error51);
@@ -29183,12 +29183,12 @@ var Protocol = class {
       }
     } catch {
     }
-    return new Promise((resolve, reject) => {
+    return new Promise((resolve2, reject) => {
       if (signal.aborted) {
         reject(new McpError(ErrorCode.InvalidRequest, "Request cancelled"));
         return;
       }
-      const timeoutId = setTimeout(resolve, interval);
+      const timeoutId = setTimeout(resolve2, interval);
       signal.addEventListener("abort", () => {
         clearTimeout(timeoutId);
         reject(new McpError(ErrorCode.InvalidRequest, "Request cancelled"));
@@ -30288,7 +30288,7 @@ var McpServer = class {
     let task = createTaskResult.task;
     const pollInterval = task.pollInterval ?? 5e3;
     while (task.status !== "completed" && task.status !== "failed" && task.status !== "cancelled") {
-      await new Promise((resolve) => setTimeout(resolve, pollInterval));
+      await new Promise((resolve2) => setTimeout(resolve2, pollInterval));
       const updatedTask = await extra.taskStore.getTask(taskId);
       if (!updatedTask) {
         throw new McpError(ErrorCode.InternalError, `Task ${taskId} not found during polling`);
@@ -30937,12 +30937,12 @@ var StdioServerTransport = class {
     this.onclose?.();
   }
   send(message) {
-    return new Promise((resolve) => {
+    return new Promise((resolve2) => {
       const json2 = serializeMessage(message);
       if (this._stdout.write(json2)) {
-        resolve();
+        resolve2();
       } else {
-        this._stdout.once("drain", resolve);
+        this._stdout.once("drain", resolve2);
       }
     });
   }
@@ -31076,7 +31076,7 @@ var pageSchema = {
 };
 
 // src/version.ts
-var VERSION = "0.1.0";
+var VERSION = "0.3.0";
 
 // src/client.ts
 import { dirname, join } from "node:path";
@@ -31099,6 +31099,7 @@ function filterImageModels(raw) {
 var __dirname = dirname(fileURLToPath(import.meta.url));
 await loadDotenvSafely({ path: join(__dirname, "..", ".env"), override: false });
 var BASE_URL = "https://generativelanguage.googleapis.com/v1beta";
+var INTERACTIONS_URL = "https://generativelanguage.googleapis.com/v1beta/interactions";
 var SERVICE = "Gemini";
 var GeminiClient = class {
   apiKey;
@@ -31148,31 +31149,121 @@ var GeminiClient = class {
     for (const img of opts.images ?? []) {
       parts.push({ inline_data: { mime_type: img.mimeType, data: img.base64 } });
     }
-    const generationConfig = { responseModalities: ["IMAGE"] };
+    if (opts.videoUrl) {
+      parts.push({ file_data: { file_uri: opts.videoUrl, mime_type: "video/mp4" } });
+    }
+    const generationConfig = { responseModalities: ["TEXT", "IMAGE"] };
     if (opts.aspectRatio || opts.imageSize) {
       const imageConfig = {};
       if (opts.aspectRatio) imageConfig.aspectRatio = opts.aspectRatio;
       if (opts.imageSize) imageConfig.imageSize = opts.imageSize;
       generationConfig.imageConfig = imageConfig;
     }
+    if (opts.seed !== void 0) generationConfig.seed = opts.seed;
+    if (opts.thinkingLevel !== void 0) generationConfig.thinkingConfig = { thinkingLevel: opts.thinkingLevel };
+    const requestBody = { contents: [{ parts }], generationConfig };
+    if (opts.googleSearch) requestBody.tools = [{ google_search: {} }];
     const data = await this.call(
       "POST",
       `/models/${model}:generateContent`,
-      { contents: [{ parts }], generationConfig }
+      requestBody
     );
-    const out = [];
+    const images = [];
+    const textParts = [];
+    let groundingMeta;
     for (const cand of data.candidates ?? []) {
       for (const part of cand.content?.parts ?? []) {
         const inline = part.inline_data ?? part.inlineData;
-        if (inline?.data) out.push({ base64: inline.data, mimeType: inline.mime_type ?? inline.mimeType ?? "image/jpeg" });
+        if (inline?.data) {
+          images.push({ base64: inline.data, mimeType: inline.mime_type ?? inline.mimeType ?? "image/jpeg" });
+        } else if (typeof part.text === "string" && part.text.trim()) {
+          textParts.push(part.text);
+        }
       }
+      if (cand.groundingMetadata) groundingMeta = cand.groundingMetadata;
     }
-    if (out.length === 0) {
+    if (images.length === 0) {
       throw new McpToolError(`${SERVICE} returned no image`, {
         hint: "The request may have been blocked by safety filters \u2014 try rephrasing the prompt."
       });
     }
-    return out;
+    const text = textParts.join("\n") || void 0;
+    let grounding;
+    if (groundingMeta) {
+      const queries = (groundingMeta.webSearchQueries ?? []).filter(Boolean);
+      const sources = (groundingMeta.groundingChunks ?? []).map((c) => c.web).filter((w) => !!w);
+      if (queries.length > 0 || sources.length > 0) {
+        grounding = {};
+        if (queries.length > 0) grounding.queries = queries;
+        if (sources.length > 0) grounding.sources = sources;
+      }
+    }
+    return { images, text, grounding };
+  }
+  async interact(opts) {
+    const key = this.requireKey();
+    const model = resolveModel(opts.model, readEnvVar("GEMINI_IMAGE_MODEL"));
+    const inputParts = [{ type: "text", text: opts.input }];
+    for (const img of opts.images ?? []) {
+      inputParts.push({ type: "image", mime_type: img.mimeType, data: img.base64 });
+    }
+    if (opts.videoUrl) {
+      inputParts.push({ type: "video", uri: opts.videoUrl, mime_type: "video/mp4" });
+    }
+    const responseFormat = { type: "image", mime_type: "image/jpeg" };
+    if (opts.aspectRatio) responseFormat.aspect_ratio = opts.aspectRatio;
+    if (opts.imageSize) responseFormat.image_size = opts.imageSize;
+    const body = { model, input: inputParts, response_format: responseFormat };
+    if (opts.thinkingLevel !== void 0) body.generation_config = { thinking_level: opts.thinkingLevel };
+    if (opts.previousInteractionId !== void 0) body.previous_interaction_id = opts.previousInteractionId;
+    if (opts.googleSearch) body.tools = [{ type: "google_search" }];
+    const res = await this.fetchImpl(INTERACTIONS_URL, {
+      method: "POST",
+      headers: {
+        "x-goog-api-key": key,
+        "content-type": "application/json",
+        "Api-Revision": "2026-05-20"
+      },
+      body: JSON.stringify(body)
+    });
+    if (!res.ok) {
+      const text = await res.text().catch(() => "");
+      let message = text;
+      try {
+        const parsed = JSON.parse(text);
+        if (parsed.error?.message) message = parsed.error.message;
+      } catch {
+      }
+      throw new McpToolError(`Gemini Interactions API ${res.status}: ${truncateErrorMessage(message)}`);
+    }
+    const data = await res.json();
+    const images = [];
+    const textParts = [];
+    const searchQueries = [];
+    for (const step of data.steps ?? []) {
+      if (step.type === "google_search_call") {
+        for (const q of step.arguments?.queries ?? []) if (q?.trim()) searchQueries.push(q);
+        continue;
+      }
+      if (step.type !== "model_output") continue;
+      for (const parts of [step.content ?? [], step.summary ?? []]) {
+        for (const part of parts) {
+          if (part.type === "image" && part.data) {
+            images.push({ base64: part.data, mimeType: part.mime_type ?? "image/jpeg" });
+          } else if (part.type === "text" && part.text?.trim()) {
+            textParts.push(part.text);
+          }
+        }
+      }
+    }
+    if (images.length === 0) {
+      throw new McpToolError("Gemini returned no image", {
+        hint: "The request may have been blocked by safety filters \u2014 try rephrasing the prompt."
+      });
+    }
+    const resultText = textParts.join("\n") || void 0;
+    const grounding = searchQueries.length > 0 ? { queries: searchQueries } : void 0;
+    return { id: data.id, images, text: resultText, grounding };
   }
 };
 var client = new GeminiClient();
@@ -31194,7 +31285,7 @@ function registerModelTools(server) {
 
 // src/images.ts
 import { mkdir, readFile, writeFile, access } from "node:fs/promises";
-import { join as join2 } from "node:path";
+import { join as join2, resolve } from "node:path";
 function slugify2(text, max = 40) {
   const s = text.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "").slice(0, max).replace(/-+$/g, "");
   return s || "image";
@@ -31216,43 +31307,93 @@ async function uniquePath(dir, base, ext) {
   }
   return candidate;
 }
+function sniffMimeBytes(buf) {
+  if (buf.length >= 4 && buf[0] === 137 && buf[1] === 80 && buf[2] === 78 && buf[3] === 71) {
+    return "image/png";
+  }
+  if (buf.length >= 3 && buf[0] === 255 && buf[1] === 216 && buf[2] === 255) {
+    return "image/jpeg";
+  }
+  if (buf.length >= 12 && buf[0] === 82 && buf[1] === 73 && buf[2] === 70 && buf[3] === 70 && buf[8] === 87 && buf[9] === 69 && buf[10] === 66 && buf[11] === 80) {
+    return "image/webp";
+  }
+  return "image/png";
+}
 async function writeImage(dir, base, base643, mimeType) {
   await mkdir(dir, { recursive: true });
-  const ext = mimeType.includes("jpeg") ? "jpg" : "png";
+  const ext = mimeType.includes("jpeg") ? "jpg" : mimeType.includes("webp") ? "webp" : "png";
   const path = await uniquePath(dir, base, ext);
   await writeFile(path, Buffer.from(base643, "base64"));
-  return path;
+  return resolve(path);
 }
 async function readImageAsInline(path) {
   const buf = await readFile(path);
-  const lower = path.toLowerCase();
-  const mimeType = lower.endsWith(".jpg") || lower.endsWith(".jpeg") ? "image/jpeg" : "image/png";
+  const mimeType = sniffMimeBytes(buf);
   return { base64: buf.toString("base64"), mimeType };
 }
+function decodeImageInput(input) {
+  const trimmed = input.trim();
+  if (trimmed.startsWith("data:")) {
+    const marker = trimmed.indexOf(";base64,");
+    if (marker === -1) {
+      throw new Error(`Unsupported data URI (expected ;base64,<data>): ${trimmed.slice(0, 48)}\u2026`);
+    }
+    const mimeType = trimmed.slice(5, trimmed.indexOf(";", 5)) || "image/png";
+    return { mimeType, base64: trimmed.slice(marker + ";base64,".length) };
+  }
+  const buf = Buffer.from(trimmed, "base64");
+  return { base64: trimmed, mimeType: sniffMimeBytes(buf) };
+}
+async function loadImageInputs(paths = [], base64s = []) {
+  const fromPaths = await Promise.all(paths.map((p) => readImageAsInline(p)));
+  const fromBase64s = base64s.map((b) => decodeImageInput(b));
+  return [...fromPaths, ...fromBase64s];
+}
+function baseName(name) {
+  const stripped = name.replace(/\.(png|jpe?g|webp)$/i, "");
+  const safe = stripped.replace(/[^A-Za-z0-9._-]+/g, "-").replace(/-+/g, "-").replace(/^-+|-+$/g, "");
+  return safe || "image";
+}
 function resolveOutputDir(perCall) {
   return perCall?.trim() || readEnvVar("GEMINI_OUTPUT_DIR") || process.cwd();
 }
 
 // src/tools/shared.ts
 var ASPECT_RATIOS = ["1:1", "2:3", "3:2", "3:4", "4:3", "4:5", "5:4", "9:16", "16:9", "21:9", "1:4", "4:1", "1:8", "8:1"];
-var IMAGE_SIZES = ["1K", "2K", "4K"];
+var IMAGE_SIZES = ["512", "1K", "2K", "4K"];
 var sharedImageSchema = {
   model: external_exports.string().optional().describe("Model id override (default: server default; see gemini_list_models)"),
   aspect_ratio: external_exports.enum(ASPECT_RATIOS).optional().describe("Output aspect ratio"),
-  image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution"),
+  image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution (512 = 0.5K, Flash-only)"),
   output_dir: external_exports.string().optional().describe("Directory to write images to (default: $GEMINI_OUTPUT_DIR or cwd)"),
-  inline: external_exports.boolean().optional().describe("Return base64 images inline instead of writing to disk")
+  inline: external_exports.boolean().optional().describe("Return base64 images inline instead of writing to disk"),
+  seed: external_exports.number().int().optional().describe("Seed for reproducible generation; random if omitted"),
+  thinking_level: external_exports.enum(["minimal", "high"]).optional().describe("Reasoning depth (Gemini 3 models); higher can help complex/structural edits"),
+  google_search: external_exports.boolean().optional().describe("Ground the image in live Google Search results (current events, weather, data)")
 };
-async function emit(named, opts) {
+function pickSeed(seed) {
+  return seed ?? Math.floor(Math.random() * 2147483e3);
+}
+function buildMeta(model, seed, opts) {
+  const meta3 = { model, seed };
+  if (opts.aspect_ratio) meta3.aspect_ratio = opts.aspect_ratio;
+  if (opts.image_size) meta3.image_size = opts.image_size;
+  return meta3;
+}
+async function emit(named, opts, meta3) {
   if (opts.inline) {
-    return {
-      content: named.map((n) => ({ type: "image", data: n.image.base64, mimeType: n.image.mimeType }))
-    };
+    const imageBlocks = named.map((n) => ({ type: "image", data: n.image.base64, mimeType: n.image.mimeType }));
+    if (meta3 && Object.keys(meta3).length > 0) {
+      return {
+        content: [{ type: "text", text: JSON.stringify(meta3, null, 2) }, ...imageBlocks]
+      };
+    }
+    return { content: imageBlocks };
   }
   const dir = resolveOutputDir(opts.output_dir);
   const images = [];
   for (const n of named) images.push(await writeImage(dir, n.base, n.image.base64, n.image.mimeType));
-  return textResult({ images });
+  return textResult({ images, ...meta3 });
 }
 
 // src/tools/generate.ts
@@ -31265,46 +31406,83 @@ function registerGenerateTools(server) {
       inputSchema: {
         prompt: external_exports.string().min(1).describe("Text prompt describing the image"),
         count: external_exports.number().int().positive().max(8).optional().describe("Number of independent images (default 1)"),
+        filename: external_exports.string().optional().describe("Base filename for the output image (extension stripped; default: slugified prompt)"),
+        images: external_exports.array(external_exports.string().min(1)).optional().describe("Paths to reference input images (image-conditioned generation)"),
+        images_base64: external_exports.array(external_exports.string().min(1)).optional().describe("Reference images as base64 strings or data URIs"),
+        video_url: external_exports.string().url().optional().describe("Public YouTube URL as a video reference (video\u2192image; use a Flash model e.g. gemini-3.1-flash-image)"),
         ...sharedImageSchema
       }
     },
     async (args) => {
       const count = args.count ?? 1;
-      const slug = slugify2(args.prompt);
+      const seed = pickSeed(args.seed);
+      const model = resolveModel(args.model, readEnvVar("GEMINI_IMAGE_MODEL"));
+      const slug = args.filename ? baseName(args.filename) : slugify2(args.prompt);
+      const refInputs = await loadImageInputs(args.images, args.images_base64);
       const named = [];
+      let capturedText;
+      let capturedGrounding;
       for (let i = 0; i < count; i++) {
-        const [img] = await client.generate({
+        const result = await client.generate({
           prompt: args.prompt,
+          images: refInputs.length > 0 ? refInputs : void 0,
           model: args.model,
           aspectRatio: args.aspect_ratio,
-          imageSize: args.image_size
+          imageSize: args.image_size,
+          seed: seed + i,
+          thinkingLevel: args.thinking_level,
+          googleSearch: args.google_search,
+          videoUrl: args.video_url
         });
+        const { images: [img], text } = result;
+        if (text && !capturedText) capturedText = text;
+        if (result.grounding && !capturedGrounding) capturedGrounding = result.grounding;
         named.push({ image: img, base: count === 1 ? slug : `${slug}-${String(i + 1).padStart(2, "0")}` });
       }
-      return emit(named, args);
+      const meta3 = buildMeta(model, seed, args);
+      if (capturedText) meta3.text = capturedText;
+      if (capturedGrounding) meta3.grounding = capturedGrounding;
+      return emit(named, args, meta3);
     }
   );
   server.registerTool(
     "gemini_edit_image",
     {
-      description: "Edit or compose images: provide one image to edit, or multiple images to combine/blend, plus a text instruction.",
+      description: "Edit or compose images: provide one or more input images (paths or base64), plus a text instruction. Gemini over-preserves the input; there is no edit-strength control \u2014 for large structural changes, reroll with a different `seed` or more forceful wording.",
       annotations: { readOnlyHint: false, openWorldHint: true },
       inputSchema: {
         prompt: external_exports.string().min(1).describe("Instruction describing the edit or composition"),
-        images: external_exports.array(external_exports.string().min(1)).min(1).describe("Paths to input image file(s) (1 = edit, 2+ = compose)"),
+        images: external_exports.array(external_exports.string().min(1)).optional().describe("Paths to input image file(s) (1 = edit, 2+ = compose)"),
+        images_base64: external_exports.array(external_exports.string().min(1)).optional().describe("Input images as base64 strings or data URIs"),
+        filename: external_exports.string().optional().describe("Base filename for the output image (extension stripped; default: slugified prompt)"),
         ...sharedImageSchema
       }
     },
     async (args) => {
-      const inputs = await Promise.all(args.images.map((p) => readImageAsInline(p)));
-      const [img] = await client.generate({
+      const hasPaths = (args.images?.length ?? 0) > 0;
+      const hasBase64 = (args.images_base64?.length ?? 0) > 0;
+      if (!hasPaths && !hasBase64) {
+        throw new McpToolError("Provide at least one input image via `images` or `images_base64`.");
+      }
+      const seed = pickSeed(args.seed);
+      const model = resolveModel(args.model, readEnvVar("GEMINI_IMAGE_MODEL"));
+      const slug = args.filename ? baseName(args.filename) : slugify2(args.prompt);
+      const inputs = await loadImageInputs(args.images, args.images_base64);
+      const result = await client.generate({
         prompt: args.prompt,
         images: inputs,
         model: args.model,
         aspectRatio: args.aspect_ratio,
-        imageSize: args.image_size
+        imageSize: args.image_size,
+        seed,
+        thinkingLevel: args.thinking_level,
+        googleSearch: args.google_search
       });
-      return emit([{ image: img, base: slugify2(args.prompt) }], args);
+      const { images: [img], text } = result;
+      const meta3 = buildMeta(model, seed, args);
+      if (text) meta3.text = text;
+      if (result.grounding) meta3.grounding = result.grounding;
+      return emit([{ image: img, base: slug }], args, meta3);
     }
   );
 }
@@ -31321,6 +31499,9 @@ function registerSetTools(server) {
         scenes: external_exports.array(external_exports.string().min(1)).min(1).max(8).optional().describe("Per-image prompts (1-8); each references the master"),
         count: external_exports.number().int().positive().max(8).optional().describe("Number of variations of master_prompt (when scenes omitted)"),
         reference_mode: external_exports.enum(["master", "chain"]).optional().describe("master: every image references the master (default). chain: each references the previous."),
+        basename: external_exports.string().optional().describe("Base filename prefix for output images (default: slugified master_prompt)"),
+        master_images: external_exports.array(external_exports.string().min(1)).optional().describe("Reference image paths passed to the master generation call"),
+        master_images_base64: external_exports.array(external_exports.string().min(1)).optional().describe("Reference images as base64 strings or data URIs for master generation"),
         ...sharedImageSchema
       }
     },
@@ -31328,26 +31509,88 @@ function registerSetTools(server) {
       if (args.scenes && args.count || !args.scenes && !args.count) {
         throw new McpToolError("Provide exactly one of `scenes` or `count`.");
       }
-      const cfg = { model: args.model, aspectRatio: args.aspect_ratio, imageSize: args.image_size };
-      const slug = slugify2(args.master_prompt);
-      const [master] = await client.generate({ prompt: args.master_prompt, ...cfg });
+      const seed = pickSeed(args.seed);
+      const model = resolveModel(args.model, readEnvVar("GEMINI_IMAGE_MODEL"));
+      const cfg = { model: args.model, aspectRatio: args.aspect_ratio, imageSize: args.image_size, thinkingLevel: args.thinking_level, googleSearch: args.google_search };
+      const slug = args.basename ? baseName(args.basename) : slugify2(args.master_prompt);
+      const masterRefInputs = await loadImageInputs(args.master_images, args.master_images_base64);
+      const masterResult = await client.generate({
+        prompt: args.master_prompt,
+        images: masterRefInputs.length > 0 ? masterRefInputs : void 0,
+        seed,
+        ...cfg
+      });
+      const { images: [master], text: masterText } = masterResult;
       const named = [{ image: master, base: `${slug}-master` }];
       const scenePrompts = args.scenes ?? Array.from({ length: args.count ?? 0 }, () => args.master_prompt);
       const mode = args.reference_mode ?? "master";
       if (mode === "chain") {
         let ref = master;
         for (let i = 0; i < scenePrompts.length; i++) {
-          const [img] = await client.generate({ prompt: scenePrompts[i], images: [ref], ...cfg });
+          const { images: [img] } = await client.generate({ prompt: scenePrompts[i], images: [ref], seed: seed + i + 1, ...cfg });
           named.push({ image: img, base: `${slug}-${String(i + 1).padStart(2, "0")}` });
           ref = img;
         }
       } else {
         const scenes = await Promise.all(
-          scenePrompts.map((p) => client.generate({ prompt: p, images: [master], ...cfg }).then((r) => r[0]))
+          scenePrompts.map((p, i) => client.generate({ prompt: p, images: [master], seed: seed + i + 1, ...cfg }).then((r) => r.images[0]))
         );
         scenes.forEach((img, i) => named.push({ image: img, base: `${slug}-${String(i + 1).padStart(2, "0")}` }));
       }
-      return emit(named, args);
+      const meta3 = buildMeta(model, seed, args);
+      if (masterText) meta3.text = masterText;
+      if (masterResult.grounding) meta3.grounding = masterResult.grounding;
+      return emit(named, args, meta3);
+    }
+  );
+}
+
+// src/tools/interact.ts
+function registerInteractTools(server) {
+  server.registerTool(
+    "gemini_interact",
+    {
+      description: "Multi-turn image generation/editing via Gemini's Interactions API (Beta). Returns an interaction `id`; pass it back as `previous_interaction_id` to iteratively refine the SAME image conversationally (the recommended way to make incremental edits). Output is JPEG.",
+      annotations: { readOnlyHint: false, openWorldHint: true },
+      inputSchema: {
+        input: external_exports.string().min(1).describe("Text prompt or editing instruction"),
+        previous_interaction_id: external_exports.string().optional().describe("ID from a prior gemini_interact call \u2014 continues that multi-turn conversation"),
+        images: external_exports.array(external_exports.string().min(1)).optional().describe("Paths to reference input images"),
+        images_base64: external_exports.array(external_exports.string().min(1)).optional().describe("Reference images as base64 strings or data URIs"),
+        model: external_exports.string().optional().describe("Model id override (default: server default; see gemini_list_models)"),
+        aspect_ratio: external_exports.enum(ASPECT_RATIOS).optional().describe("Output aspect ratio"),
+        image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution (512 = 0.5K, Flash-only)"),
+        thinking_level: external_exports.enum(["minimal", "high"]).optional().describe("Reasoning depth; higher can help complex/structural edits"),
+        filename: external_exports.string().optional().describe("Base filename for the output image (extension stripped; default: slugified input)"),
+        output_dir: external_exports.string().optional().describe("Directory to write images to (default: $GEMINI_OUTPUT_DIR or cwd)"),
+        inline: external_exports.boolean().optional().describe("Return base64 images inline instead of writing to disk"),
+        google_search: external_exports.boolean().optional().describe("Ground the image in live Google Search results (current events, weather, data)"),
+        video_url: external_exports.string().url().optional().describe("Public YouTube URL as a video reference (video\u2192image; use a Flash model e.g. gemini-3.1-flash-image)")
+      }
+    },
+    async (args) => {
+      const inputs = await loadImageInputs(args.images, args.images_base64);
+      const r = await client.interact({
+        input: args.input,
+        images: inputs.length ? inputs : void 0,
+        model: args.model,
+        aspectRatio: args.aspect_ratio,
+        imageSize: args.image_size,
+        thinkingLevel: args.thinking_level,
+        previousInteractionId: args.previous_interaction_id,
+        googleSearch: args.google_search,
+        videoUrl: args.video_url
+      });
+      const model = resolveModel(args.model, readEnvVar("GEMINI_IMAGE_MODEL"));
+      const meta3 = { model, interaction_id: r.id };
+      if (r.text) meta3.text = r.text;
+      if (r.grounding) meta3.grounding = r.grounding;
+      const slug = args.filename ? baseName(args.filename) : slugify2(args.input);
+      const named = r.images.map((image, i) => ({
+        image,
+        base: r.images.length > 1 ? `${slug}-${String(i + 1).padStart(2, "0")}` : slug
+      }));
+      return emit(named, args, meta3);
     }
   );
 }
@@ -31357,5 +31600,5 @@ await runMcp({
   name: "gemini-mcp",
   version: VERSION,
   banner: "[gemini-mcp] This project was developed and is maintained by AI (Claude). Use at your own discretion.",
-  tools: [registerModelTools, registerGenerateTools, registerSetTools]
+  tools: [registerModelTools, registerGenerateTools, registerSetTools, registerInteractTools]
 });