@chrischall/gemini-mcp 0.7.1 → 0.8.0

package/dist/bundle.js

@@ -32106,7 +32106,7 @@ var init_server3 = __esm({
 
 // node_modules/@chrischall/mcp-utils/dist/errors/index.js
 function redactSecrets(text) {
-  return text.replace(BEARER_RE, "$1[REDACTED]").replace(BASIC_AUTH_RE, "$1[REDACTED]").replace(SET_COOKIE_RE, "$1$2=[REDACTED]").replace(COOKIE_HEADER_RE, (_m, prefix, pairs) => `${prefix}${pairs.replace(/=[^;,\s]*/g, "=[REDACTED]")}`).replace(API_KEY_RE, "[REDACTED]").replace(QUERY_SECRET_RE, "$1[REDACTED]").replace(JWT_RE, "[REDACTED]");
+  return text.replace(BEARER_RE, "$1[REDACTED]").replace(BASIC_AUTH_RE, "$1[REDACTED]").replace(SET_COOKIE_RE, "$1$2=[REDACTED]").replace(COOKIE_HEADER_RE, (_m, prefix, pairs) => `${prefix}${pairs.replace(/=[^;,\s]*/g, "=[REDACTED]")}`).replace(API_KEY_RE, "[REDACTED]").replace(QUERY_SECRET_RE, "$1[REDACTED]").replace(AWS_SIGV4_RE, "$1[REDACTED]").replace(JSON_SECRET_DQ_RE, "$1[REDACTED]$2").replace(JSON_SECRET_SQ_RE, "$1[REDACTED]$2").replace(JWT_RE, "[REDACTED]");
 }
 function truncateErrorMessage(text, max = DEFAULT_ERROR_MESSAGE_MAX) {
   const str = text === null || text === void 0 ? "" : String(text);
@@ -32115,7 +32115,7 @@ function truncateErrorMessage(text, max = DEFAULT_ERROR_MESSAGE_MAX) {
     return redacted;
   return `${redacted.slice(0, max)}\u2026 [truncated]`;
 }
-var DEFAULT_ERROR_MESSAGE_MAX, McpToolError, BEARER_RE, JWT_RE, BASIC_AUTH_RE, SET_COOKIE_RE, COOKIE_HEADER_RE, API_KEY_RE, QUERY_SECRET_RE;
+var DEFAULT_ERROR_MESSAGE_MAX, McpToolError, BEARER_RE, JWT_RE, BASIC_AUTH_RE, SET_COOKIE_RE, COOKIE_HEADER_RE, API_KEY_RE, QUERY_SECRET_RE, AWS_SIGV4_RE, JSON_SECRET_KEYS, JSON_SECRET_DQ_RE, JSON_SECRET_SQ_RE;
 var init_errors4 = __esm({
   "node_modules/@chrischall/mcp-utils/dist/errors/index.js"() {
     DEFAULT_ERROR_MESSAGE_MAX = 500;
@@ -32150,6 +32150,10 @@ var init_errors4 = __esm({
       // webhook signing secret (Stripe-style)
     ].map((p) => `\\b${p}`).join("|"), "g");
     QUERY_SECRET_RE = /([?&](?:access_token|refresh_token|client_secret|api_?key|signature|token|key|sig)=)[^&#\s"'<>`]+/gi;
+    AWS_SIGV4_RE = /([?&]X-Amz-(?:Signature|Security-Token|Credential)=)[^&#\s"'<>`]+/gi;
+    JSON_SECRET_KEYS = "access_token|refresh_token|client_secret|api_?key|password|passwd|secret|token";
+    JSON_SECRET_DQ_RE = new RegExp(`("(?:${JSON_SECRET_KEYS})"\\s*:\\s*")[^"]*(")`, "gi");
+    JSON_SECRET_SQ_RE = new RegExp(`('(?:${JSON_SECRET_KEYS})'\\s*:\\s*')[^']*(')`, "gi");
   }
 });
 
@@ -32200,6 +32204,13 @@ var init_config = __esm({
   }
 });
 
+// node_modules/@chrischall/mcp-utils/dist/fs/output.js
+var init_output = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/fs/output.js"() {
+    init_config();
+  }
+});
+
 // node_modules/@chrischall/mcp-utils/dist/fs/index.js
 import { openAsBlob } from "node:fs";
 async function fileBlob(path, opts = {}) {
@@ -32216,6 +32227,7 @@ async function fileBlob(path, opts = {}) {
 }
 var init_fs = __esm({
   "node_modules/@chrischall/mcp-utils/dist/fs/index.js"() {
+    init_output();
   }
 });
 
@@ -32225,6 +32237,28 @@ var init_throttle = __esm({
   }
 });
 
+// node_modules/@chrischall/mcp-utils/dist/http/response-cache.js
+var init_response_cache = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/http/response-cache.js"() {
+  }
+});
+
+// node_modules/@chrischall/mcp-utils/dist/http/net-atoms.js
+function parseRetryAfterMs(header, opts = {}) {
+  const defaultMs = opts.defaultMs ?? 2e3;
+  const capMs = opts.capMs ?? 3e4;
+  if (header == null)
+    return defaultMs;
+  const trimmed = header.trim();
+  if (!/^\d+$/.test(trimmed))
+    return defaultMs;
+  return Math.min(Number(trimmed) * 1e3, capMs);
+}
+var init_net_atoms = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/http/net-atoms.js"() {
+  }
+});
+
 // node_modules/@chrischall/mcp-utils/dist/http/index.js
 function hostOf(baseUrl) {
   try {
@@ -32280,7 +32314,11 @@ function createApiClient(opts) {
       const res = await once();
       if (retryStatuses.includes(res.status) && attempt < retry.count) {
         attempt += 1;
-        await sleep(retry.delayMs);
+        const delay = retry.honorRetryAfter ? parseRetryAfterMs(res.headers.get("retry-after"), {
+          defaultMs: retry.delayMs,
+          capMs: retry.maxRetryAfterMs ?? 3e4
+        }) : retry.delayMs;
+        await sleep(delay);
         continue;
       }
       return res;
@@ -32315,7 +32353,26 @@ function createApiClient(opts) {
     }
     return text;
   }
-  return { fetchJson, fetchHtml };
+  async function fetchRaw(method, path, opt = {}) {
+    const headers = { Accept: "*/*", ...opt.headers };
+    const res = await send(method, path, { ...opt, headers });
+    if (res.status === 401)
+      throw unauthorized();
+    if (res.status === 429)
+      throw rateLimited();
+    if (!res.ok) {
+      const text = await res.text().catch(() => "");
+      throw new ApiError(res.status, formatApiError(res.status, method, path, text, { service }));
+    }
+    const bytes = new Uint8Array(await res.arrayBuffer());
+    return {
+      status: res.status,
+      contentType: res.headers.get("content-type"),
+      headers: res.headers,
+      bytes
+    };
+  }
+  return { fetchJson, fetchHtml, fetchRaw };
 }
 function buildQueryString(params) {
   const parts = [];
@@ -32345,6 +32402,9 @@ var init_http = __esm({
   "node_modules/@chrischall/mcp-utils/dist/http/index.js"() {
     init_errors4();
     init_throttle();
+    init_response_cache();
+    init_net_atoms();
+    init_net_atoms();
     DEFAULT_RETRY = { count: 1, delayMs: 2e3 };
     defaultSleep = (ms) => new Promise((r) => setTimeout(r, ms));
     UnauthorizedError = class extends Error {
@@ -32384,9 +32444,16 @@ var init_http = __esm({
   }
 });
 
+// node_modules/@chrischall/mcp-utils/dist/concurrency/single-flight.js
+var init_single_flight = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/concurrency/single-flight.js"() {
+  }
+});
+
 // node_modules/@chrischall/mcp-utils/dist/concurrency/index.js
 var init_concurrency = __esm({
   "node_modules/@chrischall/mcp-utils/dist/concurrency/index.js"() {
+    init_single_flight();
   }
 });
 
@@ -32396,11 +32463,19 @@ var init_dates = __esm({
   }
 });
 
+// node_modules/@chrischall/mcp-utils/dist/zod/parse-lenient.js
+var init_parse_lenient = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/zod/parse-lenient.js"() {
+    init_errors4();
+  }
+});
+
 // node_modules/@chrischall/mcp-utils/dist/zod/index.js
 var PositiveInt, NonNegInt, NonEmptyString, IsoDate, IsoTime, NumericIdString, SafePathSegment, schemaOrigin, schemaConfirm, paginationSchema, pageSchema;
 var init_zod2 = __esm({
   "node_modules/@chrischall/mcp-utils/dist/zod/index.js"() {
     init_zod();
+    init_parse_lenient();
     PositiveInt = external_exports.number().int().positive();
     NonNegInt = external_exports.number().int().nonnegative();
     NonEmptyString = external_exports.string().min(1);
@@ -32421,12 +32496,32 @@ var init_zod2 = __esm({
   }
 });
 
+// node_modules/@chrischall/mcp-utils/dist/auth/cached-token.js
+var init_cached_token = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/auth/cached-token.js"() {
+  }
+});
+
+// node_modules/@chrischall/mcp-utils/dist/auth/es256.js
+var init_es256 = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/auth/es256.js"() {
+  }
+});
+
 // node_modules/@chrischall/mcp-utils/dist/auth/index.js
 var init_auth = __esm({
   "node_modules/@chrischall/mcp-utils/dist/auth/index.js"() {
     init_config();
     init_errors4();
     init_http();
+    init_cached_token();
+    init_es256();
+  }
+});
+
+// node_modules/@chrischall/mcp-utils/dist/scrape/index.js
+var init_scrape = __esm({
+  "node_modules/@chrischall/mcp-utils/dist/scrape/index.js"() {
   }
 });
 
@@ -32443,6 +32538,7 @@ var init_dist = __esm({
     init_dates();
     init_zod2();
     init_auth();
+    init_scrape();
   }
 });
 
@@ -32510,7 +32606,7 @@ var init_clipboard = __esm({
 init_dist();
 
 // src/version.ts
-var VERSION = "0.7.1";
+var VERSION = "0.8.0";
 
 // src/tools/models.ts
 init_dist();
@@ -32540,6 +32636,13 @@ var BASE_URL = "https://generativelanguage.googleapis.com/v1beta";
 var SERVICE = "Gemini";
 var INTERACTIONS_SERVICE = "Gemini Interactions";
 var REQUEST_TIMEOUT_MS = 6e4;
+var REQUEST_TIMEOUT_4K_MS = 12e4;
+function resolveTimeoutMs(perCallMs, imageSize) {
+  if (perCallMs !== void 0 && perCallMs > 0) return perCallMs;
+  const env = Number(readEnvVar("GEMINI_TIMEOUT_MS"));
+  if (Number.isFinite(env) && env > 0) return env;
+  return imageSize === "4K" ? REQUEST_TIMEOUT_4K_MS : REQUEST_TIMEOUT_MS;
+}
 var UPLOAD_BASE_URL = "https://generativelanguage.googleapis.com/upload/v1beta";
 var FILE_MAX_BYTES = 2 * 1024 ** 3;
 var FILE_POLL_INTERVAL_MS = 2e3;
@@ -32549,8 +32652,7 @@ var CHAIN_404_RETRY_MS = 2e3;
 var GeminiClient = class {
   apiKey;
   configError;
-  api;
-  interactionsApi;
+  apis = /* @__PURE__ */ new Map();
   fetchImpl;
   sleep;
   constructor(opts = {}) {
@@ -32566,18 +32668,37 @@ var GeminiClient = class {
     }
     this.fetchImpl = opts.fetchImpl ?? fetch;
     this.sleep = opts.sleep ?? ((ms) => new Promise((r) => setTimeout(r, ms)));
-    const shared = {
-      tokenHeader: "x-goog-api-key",
-      getToken: () => this.requireKey(),
-      timeout: REQUEST_TIMEOUT_MS,
-      fetchImpl: this.fetchImpl
-    };
-    this.api = createApiClient({ baseUrl: BASE_URL, serviceName: SERVICE, ...shared });
-    this.interactionsApi = createApiClient({
-      baseUrl: BASE_URL,
-      serviceName: INTERACTIONS_SERVICE,
-      ...shared
-    });
+  }
+  /**
+   * The `createApiClient` pair for one timeout budget, built lazily and
+   * memoized per distinct timeout (a handful of values in practice). Both
+   * share: the key in the `x-goog-api-key` header (Gemini does not use
+   * `Authorization: Bearer`) and the default 429 retry (once after 2s — the
+   * free tier rate-limits aggressively). `getToken` calls requireKey() so key
+   * resolution stays at REQUEST time: the server boots without a key and the
+   * actionable config error (with its aistudio hint) surfaces on the first
+   * tool call, not at startup.
+   */
+  apisFor(timeoutMs) {
+    let entry = this.apis.get(timeoutMs);
+    if (!entry) {
+      const shared = {
+        tokenHeader: "x-goog-api-key",
+        getToken: () => this.requireKey(),
+        timeout: timeoutMs,
+        fetchImpl: this.fetchImpl
+      };
+      entry = {
+        api: createApiClient({ baseUrl: BASE_URL, serviceName: SERVICE, ...shared }),
+        // Separate client for the Interactions API so errors name the right API.
+        // The beta-era Api-Revision header is gone: the API went GA and requests
+        // succeed without it (verified live 2026-07-06; see docs/GEMINI-API.md).
+        // Don't re-add a pinned revision — it would freeze us on old semantics.
+        interactionsApi: createApiClient({ baseUrl: BASE_URL, serviceName: INTERACTIONS_SERVICE, ...shared })
+      };
+      this.apis.set(timeoutMs, entry);
+    }
+    return entry;
   }
   requireKey() {
     if (this.configError) throw this.configError;
@@ -32587,8 +32708,9 @@ var GeminiClient = class {
   defaultModel() {
     return resolveModel(void 0, readEnvVar("GEMINI_IMAGE_MODEL"));
   }
-  async call(method, path, body) {
-    return this.api.fetchJson(method, path, body !== void 0 ? { body } : {});
+  async call(method, path, body, timeoutMs) {
+    const { api } = this.apisFor(timeoutMs ?? resolveTimeoutMs());
+    return api.fetchJson(method, path, body !== void 0 ? { body } : {});
   }
   /**
    * Upload a local video to the Gemini Files API (resumable protocol) and wait
@@ -32704,7 +32826,8 @@ var GeminiClient = class {
     const data = await this.call(
       "POST",
       `/models/${model}:generateContent`,
-      requestBody
+      requestBody,
+      resolveTimeoutMs(opts.timeoutMs, opts.imageSize)
     );
     const images = [];
     const textParts = [];
@@ -32758,10 +32881,11 @@ var GeminiClient = class {
       if (opts.searchTypes?.length) searchTool.search_types = opts.searchTypes;
       body.tools = [searchTool];
     }
+    const { interactionsApi } = this.apisFor(resolveTimeoutMs(opts.timeoutMs, opts.imageSize));
     let data;
     for (let attempt = 0; ; attempt++) {
       try {
-        data = await this.interactionsApi.fetchJson("POST", "/interactions", {
+        data = await interactionsApi.fetchJson("POST", "/interactions", {
           body
         });
         break;
@@ -32844,7 +32968,7 @@ init_dist();
 
 // src/images.ts
 init_dist();
-import { mkdir, readFile as readFile2, writeFile, access } from "node:fs/promises";
+import { mkdir, readFile as readFile2, writeFile, access, stat, open } from "node:fs/promises";
 import { existsSync } from "node:fs";
 import { join as join3, resolve, isAbsolute } from "node:path";
 function slugify2(text, max = 40) {
@@ -32887,12 +33011,15 @@ async function writeImage(dir, base, base643, mimeType) {
   await writeFile(path, Buffer.from(base643, "base64"));
   return resolve(path);
 }
+async function writeSidecar(imagePath, data) {
+  await writeFile(`${imagePath}.json`, JSON.stringify(data, null, 2));
+}
 function resolveInputPath(p, kind) {
   if (isAbsolute(p) && existsSync(p)) return p;
   const inputDir = readEnvVar("GEMINI_INPUT_DIR");
   if (inputDir) {
     const candidate = join3(inputDir, p);
-    if (existsSync(candidate)) return candidate;
+    if (existsSync(candidate)) return resolve(candidate);
   }
   const cwd = resolve(p);
   if (existsSync(cwd)) return cwd;
@@ -32930,6 +33057,23 @@ function videoMimeType(p) {
   }
   return mime;
 }
+async function previewImageInput(path) {
+  const resolved = resolveImagePath(path);
+  const { size } = await stat(resolved);
+  const fh = await open(resolved, "r");
+  try {
+    const head = Buffer.alloc(16);
+    const { bytesRead } = await fh.read(head, 0, 16, 0);
+    return { path: resolved, mimeType: sniffMimeBytes(head.subarray(0, bytesRead)), size };
+  } finally {
+    await fh.close();
+  }
+}
+async function previewVideoInput(path) {
+  const resolved = resolveVideoPath(path);
+  const { size } = await stat(resolved);
+  return { path: resolved, mimeType: videoMimeType(resolved), size };
+}
 async function readImageAsInline(path) {
   const resolved = resolveImagePath(path);
   const buf = await readFile2(resolved);
@@ -32974,6 +33118,9 @@ init_dist();
 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 = ["512", "1K", "2K", "4K"];
 var MODEL_CHOICE_GUIDE = "gemini-3.1-flash-image (Nano Banana 2) is the versatile generalist workhorse \u2014 balances speed with state-of-the-art 4K generation, world knowledge, and reliable text rendering; excels at multi-reference-image processing and consistency. gemini-3-pro-image (Nano Banana Pro) is the premium choice for the most complex visual tasks \u2014 highest world knowledge, advanced localization, accurate brand consistency, precision creative control. gemini-3.1-flash-lite-image (Nano Banana 2 Lite) is the fastest/cheapest for simple tasks (1K only, no search grounding).";
+var timeoutMsSchema = external_exports.number().int().positive().optional().describe(
+  "Upstream request timeout in ms for this call (default: $GEMINI_TIMEOUT_MS, else 60000 \u2014 or 120000 when image_size is 4K, which routinely runs past 60s)"
+);
 var sharedImageSchema = {
   // Bare id only: the model is interpolated into the URL path
   // (`/models/${model}:generateContent`), so slashes/colons/queries must be
@@ -32986,7 +33133,8 @@ var sharedImageSchema = {
   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)"),
-  from_clipboard: external_exports.boolean().optional().describe("Use the image currently on the macOS system clipboard as an input (downscaled to JPEG)")
+  from_clipboard: external_exports.boolean().optional().describe("Use the image currently on the macOS system clipboard as an input (downscaled to JPEG)"),
+  timeout_ms: timeoutMsSchema
 };
 async function resolveVideoInput(args) {
   if (args.video_url && args.video_path) {
@@ -33005,13 +33153,40 @@ var videoPathSchema = external_exports.string().optional().describe(
 function pickSeed(seed) {
   return seed ?? Math.floor(Math.random() * 2147483e3);
 }
+var HEARTBEAT_DEFAULT_MS = 1e4;
+async function withProgressHeartbeat(extra, message, fn) {
+  const progressToken = extra?._meta?.progressToken;
+  const send = extra?.sendNotification;
+  const env = Number(readEnvVar("GEMINI_HEARTBEAT_MS"));
+  const intervalMs = Number.isFinite(env) ? env : HEARTBEAT_DEFAULT_MS;
+  if (progressToken === void 0 || !send || intervalMs <= 0) return fn();
+  let ticks = 0;
+  const timer = setInterval(() => {
+    ticks++;
+    void send({
+      method: "notifications/progress",
+      params: {
+        progressToken,
+        progress: ticks,
+        message: `${message} \u2014 still working (${Math.round(ticks * intervalMs / 1e3)}s)`
+      }
+    }).catch(() => {
+    });
+  }, intervalMs);
+  timer.unref?.();
+  try {
+    return await fn();
+  } finally {
+    clearInterval(timer);
+  }
+}
 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) {
+async function emit(named, opts, meta3, onWritten) {
   if (opts.inline) {
     const imageBlocks = named.map((n) => ({ type: "image", data: n.image.base64, mimeType: n.image.mimeType }));
     if (meta3 && Object.keys(meta3).length > 0) {
@@ -33024,10 +33199,35 @@ async function emit(named, opts, meta3) {
   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));
+  await onWritten?.(images);
   return textResult({ images, ...meta3 });
 }
 
+// src/tools/_confirm.ts
+init_dist();
+function previewUnlessConfirmed(confirm, action, method, path, body) {
+  if (confirm === true) return null;
+  return textResult({
+    dryRun: true,
+    action,
+    method,
+    path,
+    ...body !== void 0 ? { willSend: body } : {},
+    note: "Re-run with confirm: true to execute."
+  });
+}
+async function previewLocalInputsUnlessConfirmed(confirm, action, endpoint, imagePaths, videoPath) {
+  const hasLocal = (imagePaths?.length ?? 0) > 0 || Boolean(videoPath);
+  if (confirm === true || !hasLocal) return null;
+  const inputs = [];
+  for (const p of imagePaths ?? []) inputs.push(await previewImageInput(p));
+  if (videoPath) inputs.push(await previewVideoInput(videoPath));
+  return previewUnlessConfirmed(confirm, action, "POST", endpoint, { inputs });
+}
+
 // src/tools/generate.ts
+var GENERATE_ENDPOINT = "/v1beta/models/{model}:generateContent";
+var SEND_LOCAL_ACTION = "Send local file input(s) to the Gemini API";
 var REFINE_HINT = "To iteratively refine this image, use gemini_interact (pass the output file via `images` on the first call, then chain `previous_interaction_id` on each subsequent call).";
 function registerGenerateTools(server) {
   server.registerTool(
@@ -33043,37 +33243,43 @@ function registerGenerateTools(server) {
         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 (or a previously uploaded Files API uri) as a video reference (video\u2192image; use a Flash model e.g. gemini-3.1-flash-image)"),
         video_path: videoPathSchema,
+        confirm: schemaConfirm,
         ...sharedImageSchema
       }
     },
-    async (args) => {
+    async (args, extra) => {
+      const gate = await previewLocalInputsUnlessConfirmed(args.confirm, SEND_LOCAL_ACTION, GENERATE_ENDPOINT, args.images, args.video_path);
+      if (gate) return gate;
       const count = args.count ?? 1;
       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 gatherImageInputs({ images: args.images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
-      const video = await resolveVideoInput(args);
+      const video = await withProgressHeartbeat(extra, "Uploading video to the Gemini Files API", () => resolveVideoInput(args));
       const named = [];
       let capturedText;
       let capturedGrounding;
-      for (let i = 0; i < count; i++) {
-        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,
-          seed: seed + i,
-          thinkingLevel: args.thinking_level,
-          googleSearch: args.google_search,
-          videoUrl: video.videoUrl,
-          videoMimeType: video.videoMimeType
-        });
-        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")}` });
-      }
+      await withProgressHeartbeat(extra, `Generating ${count > 1 ? `${count} images` : "image"} (${model})`, async () => {
+        for (let i = 0; i < count; i++) {
+          const result = await client.generate({
+            prompt: args.prompt,
+            images: refInputs.length > 0 ? refInputs : void 0,
+            model: args.model,
+            aspectRatio: args.aspect_ratio,
+            imageSize: args.image_size,
+            seed: seed + i,
+            thinkingLevel: args.thinking_level,
+            googleSearch: args.google_search,
+            videoUrl: video.videoUrl,
+            videoMimeType: video.videoMimeType,
+            timeoutMs: args.timeout_ms
+          });
+          const { images: [img], text } = result;
+          if (text && !capturedText) capturedText = text;
+          if (result.grounding && !capturedGrounding) capturedGrounding = result.grounding;
+          named.push({ image: img, base: count === 1 ? slug : `${slug}-${String(i + 1).padStart(2, "0")}` });
+        }
+      });
       const meta3 = buildMeta(model, seed, args);
       if (capturedText) meta3.text = capturedText;
       if (capturedGrounding) meta3.grounding = capturedGrounding;
@@ -33092,20 +33298,23 @@ function registerGenerateTools(server) {
         images: external_exports.array(external_exports.string().min(1)).optional().describe("Paths to input image file(s) (1 = edit, 2+ = compose)"),
         images_base64: external_exports.array(external_exports.string().min(1)).optional().describe("Input images as base64 strings or data URIs"),
         filename: external_exports.string().optional().describe("Base filename for the output image (extension stripped; default: slugified prompt)"),
+        confirm: schemaConfirm,
         ...sharedImageSchema
       }
     },
-    async (args) => {
+    async (args, extra) => {
       const hasPaths = (args.images?.length ?? 0) > 0;
       const hasBase64 = (args.images_base64?.length ?? 0) > 0;
       if (!hasPaths && !hasBase64 && !args.from_clipboard) {
         throw new McpToolError("Provide at least one input image via `images`, `images_base64`, or `from_clipboard`.");
       }
+      const gate = await previewLocalInputsUnlessConfirmed(args.confirm, SEND_LOCAL_ACTION, GENERATE_ENDPOINT, args.images);
+      if (gate) return gate;
       const seed = pickSeed(args.seed);
       const model = resolveModel(args.model, readEnvVar("GEMINI_IMAGE_MODEL"));
       const slug = args.filename ? baseName(args.filename) : slugify2(args.prompt);
       const inputs = await gatherImageInputs({ images: args.images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
-      const result = await client.generate({
+      const result = await withProgressHeartbeat(extra, `Editing image (${model})`, () => client.generate({
         prompt: args.prompt,
         images: inputs,
         model: args.model,
@@ -33113,8 +33322,9 @@ function registerGenerateTools(server) {
         imageSize: args.image_size,
         seed,
         thinkingLevel: args.thinking_level,
-        googleSearch: args.google_search
-      });
+        googleSearch: args.google_search,
+        timeoutMs: args.timeout_ms
+      }));
       const { images: [img], text } = result;
       const meta3 = buildMeta(model, seed, args);
       if (text) meta3.text = text;
@@ -33142,41 +33352,49 @@ function registerSetTools(server) {
         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"),
+        confirm: schemaConfirm,
         ...sharedImageSchema
       }
     },
-    async (args) => {
+    async (args, extra) => {
       if (args.scenes && args.count || !args.scenes && !args.count) {
         throw new McpToolError("Provide exactly one of `scenes` or `count`.");
       }
+      const gate = await previewLocalInputsUnlessConfirmed(args.confirm, "Send local master image input(s) to the Gemini API", "/v1beta/models/{model}:generateContent", args.master_images);
+      if (gate) return gate;
       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 cfg = { model: args.model, aspectRatio: args.aspect_ratio, imageSize: args.image_size, thinkingLevel: args.thinking_level, googleSearch: args.google_search, timeoutMs: args.timeout_ms };
       const slug = args.basename ? baseName(args.basename) : slugify2(args.master_prompt);
       const masterRefInputs = await gatherImageInputs({ images: args.master_images, images_base64: args.master_images_base64, from_clipboard: args.from_clipboard });
-      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: `${slug}-master` }];
-      const scenePrompts = args.scenes ?? Array.from({ length: args.count ?? 0 }, () => args.master_prompt);
-      const mode = args.reference_mode ?? "master";
-      if (mode === "chain") {
-        let ref = master;
-        for (let i = 0; i < scenePrompts.length; i++) {
-          const { images: [img] } = await client.generate({ prompt: scenePrompts[i], images: [ref], seed: seed + i + 1, ...cfg });
-          named.push({ image: img, base: `${slug}-${String(i + 1).padStart(2, "0")}` });
-          ref = img;
+      const named = [];
+      const masterResult = await withProgressHeartbeat(extra, `Generating image set (${model})`, async () => {
+        const result = await client.generate({
+          prompt: args.master_prompt,
+          images: masterRefInputs.length > 0 ? masterRefInputs : void 0,
+          seed,
+          ...cfg
+        });
+        const master = result.images[0];
+        named.push({ image: master, base: `${slug}-master` });
+        const scenePrompts = args.scenes ?? Array.from({ length: args.count ?? 0 }, () => args.master_prompt);
+        const mode = args.reference_mode ?? "master";
+        if (mode === "chain") {
+          let ref = master;
+          for (let i = 0; i < scenePrompts.length; i++) {
+            const { images: [img] } = await client.generate({ prompt: scenePrompts[i], images: [ref], seed: seed + i + 1, ...cfg });
+            named.push({ image: img, base: `${slug}-${String(i + 1).padStart(2, "0")}` });
+            ref = img;
+          }
+        } else {
+          const scenes = await Promise.all(
+            scenePrompts.map((p, i) => client.generate({ prompt: p, images: [master], seed: seed + i + 1, ...cfg }).then((r) => r.images[0]))
+          );
+          scenes.forEach((img, i) => named.push({ image: img, base: `${slug}-${String(i + 1).padStart(2, "0")}` }));
         }
-      } else {
-        const scenes = await Promise.all(
-          scenePrompts.map((p, i) => client.generate({ prompt: p, images: [master], seed: seed + i + 1, ...cfg }).then((r) => r.images[0]))
-        );
-        scenes.forEach((img, i) => named.push({ image: img, base: `${slug}-${String(i + 1).padStart(2, "0")}` }));
-      }
+        return result;
+      });
+      const masterText = masterResult.text;
       const meta3 = buildMeta(model, seed, args);
       if (masterText) meta3.text = masterText;
       if (masterResult.grounding) meta3.grounding = masterResult.grounding;
@@ -33189,18 +33407,33 @@ function registerSetTools(server) {
 init_zod();
 init_dist();
 var lastInteractionId;
+var writtenOutputs = /* @__PURE__ */ new Set();
+var NEW_REFERENCES_ONLY = "NEW reference images only (e.g. a style or target photo). When chaining with previous_interaction_id, do NOT re-attach the prior turn's output \u2014 the interaction already contains it, and re-sending it anchors the model against your edit.";
+function splitReattachedOutputs(images) {
+  const kept = [];
+  const dropped = [];
+  for (const p of images) {
+    let resolved;
+    try {
+      resolved = resolveImagePath(p);
+    } catch {
+    }
+    (resolved && writtenOutputs.has(resolved) ? dropped : kept).push(p);
+  }
+  return { kept, dropped };
+}
 function registerInteractTools(server) {
   server.registerTool(
     "gemini_interact",
     {
-      description: "Preferred tool for iterative or multi-step refinement of a single image \u2014 multi-turn generation/editing via Gemini's Interactions API. To refine, capture the returned interaction `id` and pass it as `previous_interaction_id` on the next call \u2014 do NOT start a new interaction or re-upload the image for each tweak. `continue_last: true` chains from this session's most recent interaction without threading the id. Output is JPEG.",
+      description: "Preferred tool for iterative or multi-step refinement of a single image \u2014 multi-turn generation/editing via Gemini's Interactions API. To refine, capture the returned interaction `id` and pass it as `previous_interaction_id` on the next call \u2014 do NOT start a new interaction or re-upload the image for each tweak. `continue_last: true` chains from this session's most recent interaction without threading the id. If a call times out on the client side, the generation usually still completes: the image plus a `<image>.json` sidecar recording its interaction id land in the output dir, and `continue_last: true` still resumes that interaction \u2014 check the output dir before re-issuing (a re-issue is a second billable generation). 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"),
         continue_last: external_exports.boolean().optional().describe("Continue from the most recent interaction this server created (convenience for previous_interaction_id; an explicit id wins)"),
-        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"),
+        images: external_exports.array(external_exports.string().min(1)).optional().describe(`Paths to reference input images. ${NEW_REFERENCES_ONLY}`),
+        images_base64: external_exports.array(external_exports.string().min(1)).optional().describe(`Reference images as base64 strings or data URIs. ${NEW_REFERENCES_ONLY}`),
         model: external_exports.string().optional().describe(`Model id override (default: server default; see gemini_list_models). ${MODEL_CHOICE_GUIDE}`),
         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)"),
@@ -33214,10 +33447,12 @@ function registerInteractTools(server) {
         ),
         video_url: external_exports.string().url().optional().describe("Public YouTube URL (or a previously uploaded Files API uri) as a video reference (video\u2192image; use a Flash model e.g. gemini-3.1-flash-image)"),
         video_path: videoPathSchema,
-        from_clipboard: external_exports.boolean().optional().describe("Use the image currently on the macOS system clipboard as an input (downscaled to JPEG)")
+        timeout_ms: timeoutMsSchema,
+        from_clipboard: external_exports.boolean().optional().describe("Use the image currently on the macOS system clipboard as an input (downscaled to JPEG)"),
+        confirm: schemaConfirm
       }
     },
-    async (args) => {
+    async (args, extra) => {
       let previousInteractionId = args.previous_interaction_id;
       if (!previousInteractionId && args.continue_last) {
         if (!lastInteractionId) {
@@ -33227,9 +33462,16 @@ function registerInteractTools(server) {
         }
         previousInteractionId = lastInteractionId;
       }
-      const inputs = await gatherImageInputs({ images: args.images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
-      const video = await resolveVideoInput(args);
-      const r = await client.interact({
+      let images = args.images;
+      let dropped = [];
+      if (previousInteractionId && images?.length) {
+        ({ kept: images, dropped } = splitReattachedOutputs(images));
+      }
+      const gate = await previewLocalInputsUnlessConfirmed(args.confirm, "Send local file input(s) to the Gemini Interactions API", "/v1beta/interactions", images, args.video_path);
+      if (gate) return gate;
+      const inputs = await gatherImageInputs({ images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
+      const video = await withProgressHeartbeat(extra, "Uploading video to the Gemini Files API", () => resolveVideoInput(args));
+      const r = await withProgressHeartbeat(extra, "Generating image (Gemini Interactions API)", () => client.interact({
         input: args.input,
         images: inputs.length ? inputs : void 0,
         model: args.model,
@@ -33240,13 +33482,15 @@ function registerInteractTools(server) {
         googleSearch: args.google_search,
         searchTypes: args.search_types,
         videoUrl: video.videoUrl,
-        videoMimeType: video.videoMimeType
-      });
+        videoMimeType: video.videoMimeType,
+        timeoutMs: args.timeout_ms
+      }));
       lastInteractionId = r.id;
       const model = resolveModel(args.model, readEnvVar("GEMINI_IMAGE_MODEL"));
       const meta3 = { model, interaction_id: r.id };
       if (previousInteractionId) meta3.previous_interaction_id = previousInteractionId;
-      meta3.hint = `To refine this image, call gemini_interact again with previous_interaction_id: "${r.id}" (or continue_last: true).`;
+      if (dropped.length) meta3.dropped_previous_output = dropped;
+      meta3.hint = `To refine this image, call gemini_interact again with previous_interaction_id: "${r.id}" (or continue_last: true) \u2014 do NOT re-attach the returned image as an input; the interaction already contains it.`;
       if (r.text) meta3.text = r.text;
       if (r.grounding) meta3.grounding = r.grounding;
       if (video.videoFileMeta) meta3.video_file = video.videoFileMeta;
@@ -33255,7 +33499,14 @@ function registerInteractTools(server) {
         image,
         base: r.images.length > 1 ? `${slug}-${String(i + 1).padStart(2, "0")}` : slug
       }));
-      return emit(named, args, meta3);
+      return emit(named, args, meta3, async (paths) => {
+        for (const p of paths) writtenOutputs.add(p);
+        try {
+          for (const p of paths) await writeSidecar(p, { ...meta3, images: paths });
+        } catch (err) {
+          meta3.sidecar_error = err instanceof Error ? err.message : String(err);
+        }
+      });
     }
   );
 }