@chrischall/gemini-mcp 0.3.0 → 0.4.0

package/dist/clipboard.js

@@ -0,0 +1,67 @@
+import { execFile } from 'node:child_process';
+import { promisify } from 'node:util';
+import { readFile, unlink } from 'node:fs/promises';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { McpToolError } from '@chrischall/mcp-utils';
+const execFileAsync = promisify(execFile);
+const defaultRun = async (cmd, args) => {
+    await execFileAsync(cmd, args);
+};
+/**
+ * Read the macOS system clipboard image into a base64 JPEG.
+ *
+ * Steps:
+ *  1. osascript: write clipboard PNG to a temp file.
+ *  2. sips: downscale + convert to JPEG (≤2048px, avoids 16-bit/large payloads).
+ *  3. Read the JPEG → base64.
+ *
+ * Temp files are cleaned up in a `finally` block.
+ */
+export async function readClipboardImage(run = defaultRun) {
+    if (process.platform !== 'darwin') {
+        throw new McpToolError('from_clipboard is only supported on macOS', {
+            hint: 'Save the image to a file and pass it via `images` instead.',
+        });
+    }
+    // Unique temp paths under tmpdir() — OS-controlled (/var/folders/…/T or /tmp on
+    // macOS), digits+alnum id, no user input. Commands run via execFile (no shell),
+    // so the only string-interpolated value (the path, in the AppleScript literal)
+    // can't reach a shell; a quote-bearing TMPDIR would just yield a parse error,
+    // caught as "No image found" below.
+    const id = `${Date.now()}-${Math.random().toString(36).slice(2)}`;
+    const png = join(tmpdir(), `gemini-clip-${id}.png`);
+    const jpg = join(tmpdir(), `gemini-clip-${id}.jpg`);
+    try {
+        // Step 1: write clipboard PNG to temp file via osascript
+        try {
+            await run('osascript', [
+                '-e', `set f to (open for access POSIX file "${png}" with write permission)`,
+                '-e', 'write (the clipboard as «class PNGf») to f',
+                '-e', 'close access f',
+            ]);
+        }
+        catch {
+            throw new McpToolError('No image found on the clipboard', {
+                hint: 'Copy an image (⌘C) first, or pass `images`/`images_base64`.',
+            });
+        }
+        // Step 2: downscale + convert to JPEG via sips
+        try {
+            await run('sips', ['-Z', '2048', '-s', 'format', 'jpeg', png, '--out', jpg]);
+        }
+        catch {
+            throw new McpToolError('Failed to convert the clipboard image', {
+                hint: 'The clipboard may contain a non-standard image format.',
+            });
+        }
+        // Step 3: read JPEG → base64
+        const buf = await readFile(jpg);
+        return { base64: buf.toString('base64'), mimeType: 'image/jpeg' };
+    }
+    finally {
+        // Clean up both temp files (ignore errors if they don't exist)
+        await unlink(png).catch(() => undefined);
+        await unlink(jpg).catch(() => undefined);
+    }
+}

package/dist/images.js

@@ -1,6 +1,7 @@
 import { mkdir, readFile, writeFile, access } from 'node:fs/promises';
-import { join, resolve } from 'node:path';
-import { readEnvVar } from '@chrischall/mcp-utils';
+import { existsSync } from 'node:fs';
+import { join, resolve, isAbsolute } from 'node:path';
+import { readEnvVar, McpToolError } from '@chrischall/mcp-utils';
 /** URL/file-safe slug from a prompt; never empty. */
 export function slugify(text, max = 40) {
     const s = text
@@ -56,9 +57,34 @@ export async function writeImage(dir, base, base64, mimeType) {
     await writeFile(path, Buffer.from(base64, 'base64'));
     return resolve(path);
 }
+/**
+ * Resolve an image path, checking (in order):
+ *  1. Absolute path that exists → returned as-is.
+ *  2. `GEMINI_INPUT_DIR` env var is set → look for `join(inputDir, p)`.
+ *  3. Relative to cwd → `resolve(p)`.
+ * Throws a helpful `McpToolError` if none are found.
+ */
+export function resolveImagePath(p) {
+    if (isAbsolute(p) && existsSync(p))
+        return p;
+    const inputDir = readEnvVar('GEMINI_INPUT_DIR');
+    if (inputDir) {
+        // NOTE: a `../`-bearing `p` can escape inputDir. Acceptable for a local,
+        // single-user MCP — the caller already has the user's own filesystem access.
+        const candidate = join(inputDir, p);
+        if (existsSync(candidate))
+            return candidate;
+    }
+    const cwd = resolve(p);
+    if (existsSync(cwd))
+        return cwd;
+    const hint = 'Set GEMINI_INPUT_DIR to a directory containing your images, or pass an absolute path.';
+    throw new McpToolError(`Image not found: ${p}` + (inputDir ? ` (also searched GEMINI_INPUT_DIR=${inputDir})` : ''), { hint });
+}
 /** Read an image file into `{ base64, mimeType }` for an inline_data part. */
 export async function readImageAsInline(path) {
-    const buf = await readFile(path);
+    const resolved = resolveImagePath(path);
+    const buf = await readFile(resolved);
     const mimeType = sniffMimeBytes(buf);
     return { base64: buf.toString('base64'), mimeType };
 }
@@ -91,6 +117,19 @@ export async function loadImageInputs(paths = [], base64s = []) {
     const fromBase64s = base64s.map((b) => decodeImageInput(b));
     return [...fromPaths, ...fromBase64s];
 }
+/**
+ * Aggregate image inputs from all sources: clipboard (macOS only), file paths, and base64 strings.
+ * Clipboard image is prepended if `from_clipboard` is true.
+ */
+export async function gatherImageInputs(opts) {
+    // Only load the clipboard module when actually requested (keeps the child_process
+    // path off the default code path).
+    const clipboardImages = opts.from_clipboard
+        ? [await (await import('./clipboard.js')).readClipboardImage()]
+        : [];
+    const fileImages = await loadImageInputs(opts.images, opts.images_base64);
+    return [...clipboardImages, ...fileImages];
+}
 /** Caller-supplied filename → safe base name (no extension). */
 export function baseName(name) {
     // Strip known image extensions

package/dist/tools/generate.js

@@ -2,7 +2,7 @@ import { z } from 'zod';
 import { McpToolError, readEnvVar } from '@chrischall/mcp-utils';
 import { resolveModel } from '../models.js';
 import { client } from '../client.js';
-import { slugify, baseName, loadImageInputs } from '../images.js';
+import { slugify, baseName, gatherImageInputs } from '../images.js';
 import { emit, sharedImageSchema, pickSeed, buildMeta } from './shared.js';
 export function registerGenerateTools(server) {
     server.registerTool('gemini_generate_image', {
@@ -22,7 +22,7 @@ export function registerGenerateTools(server) {
         const seed = pickSeed(args.seed);
         const model = resolveModel(args.model, readEnvVar('GEMINI_IMAGE_MODEL'));
         const slug = args.filename ? baseName(args.filename) : slugify(args.prompt);
-        const refInputs = await loadImageInputs(args.images, args.images_base64);
+        const refInputs = await gatherImageInputs({ images: args.images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
         const named = [];
         let capturedText;
         // For count>1, surface the FIRST call's grounding (each call grounds
@@ -70,13 +70,13 @@ export function registerGenerateTools(server) {
     }, async (args) => {
         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`.');
+        if (!hasPaths && !hasBase64 && !args.from_clipboard) {
+            throw new McpToolError('Provide at least one input image via `images`, `images_base64`, or `from_clipboard`.');
         }
         const seed = pickSeed(args.seed);
         const model = resolveModel(args.model, readEnvVar('GEMINI_IMAGE_MODEL'));
         const slug = args.filename ? baseName(args.filename) : slugify(args.prompt);
-        const inputs = await loadImageInputs(args.images, args.images_base64);
+        const inputs = await gatherImageInputs({ images: args.images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
         const result = await client.generate({
             prompt: args.prompt,
             images: inputs,

package/dist/tools/interact.js

@@ -2,7 +2,7 @@ import { z } from 'zod';
 import { readEnvVar } from '@chrischall/mcp-utils';
 import { resolveModel } from '../models.js';
 import { client } from '../client.js';
-import { slugify, baseName, loadImageInputs } from '../images.js';
+import { slugify, baseName, gatherImageInputs } from '../images.js';
 import { emit, ASPECT_RATIOS, IMAGE_SIZES } from './shared.js';
 export function registerInteractTools(server) {
     server.registerTool('gemini_interact', {
@@ -59,9 +59,13 @@ export function registerInteractTools(server) {
                 .url()
                 .optional()
                 .describe('Public YouTube URL as a video reference (video→image; use a Flash model e.g. gemini-3.1-flash-image)'),
+            from_clipboard: z
+                .boolean()
+                .optional()
+                .describe('Use the image currently on the macOS system clipboard as an input (downscaled to JPEG)'),
         },
     }, async (args) => {
-        const inputs = await loadImageInputs(args.images, args.images_base64);
+        const inputs = await gatherImageInputs({ images: args.images, images_base64: args.images_base64, from_clipboard: args.from_clipboard });
         const r = await client.interact({
             input: args.input,
             images: inputs.length ? inputs : undefined,

package/dist/tools/set.js

@@ -2,7 +2,7 @@ import { z } from 'zod';
 import { McpToolError, readEnvVar } from '@chrischall/mcp-utils';
 import { resolveModel } from '../models.js';
 import { client } from '../client.js';
-import { slugify, baseName, loadImageInputs } from '../images.js';
+import { slugify, baseName, gatherImageInputs } from '../images.js';
 import { emit, sharedImageSchema, pickSeed, buildMeta } from './shared.js';
 export function registerSetTools(server) {
     server.registerTool('gemini_generate_set', {
@@ -28,7 +28,7 @@ export function registerSetTools(server) {
         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) : slugify(args.master_prompt);
         // 1. master (optionally seeded from reference images)
-        const masterRefInputs = await loadImageInputs(args.master_images, args.master_images_base64);
+        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 : undefined,

package/dist/tools/shared.js

@@ -19,6 +19,7 @@ export const sharedImageSchema = {
     seed: z.number().int().optional().describe('Seed for reproducible generation; random if omitted'),
     thinking_level: z.enum(['minimal', 'high']).optional().describe('Reasoning depth (Gemini 3 models); higher can help complex/structural edits'),
     google_search: z.boolean().optional().describe('Ground the image in live Google Search results (current events, weather, data)'),
+    from_clipboard: z.boolean().optional().describe('Use the image currently on the macOS system clipboard as an input (downscaled to JPEG)'),
 };
 /**
  * Return the provided seed, or pick a random one. Capped well below INT32_MAX

package/dist/version.js

@@ -1,2 +1,2 @@
 /** Single source of the server version. release-please bumps the literal below. */
-export const VERSION = '0.3.0'; // x-release-please-version
+export const VERSION = '0.4.0'; // x-release-please-version

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@chrischall/gemini-mcp",
-  "version": "0.3.0",
+  "version": "0.4.0",
   "mcpName": "io.github.chrischall/gemini-mcp",
   "description": "Gemini image-generation MCP server for Claude — developed and maintained by AI (Claude Code)",
   "author": "Claude Code (AI) <https://www.anthropic.com/claude>",

package/server.json

@@ -6,12 +6,12 @@
     "url": "https://github.com/chrischall/gemini-mcp",
     "source": "github"
   },
-  "version": "0.3.0",
+  "version": "0.4.0",
   "packages": [
     {
       "registryType": "npm",
       "identifier": "@chrischall/gemini-mcp",
-      "version": "0.3.0",
+      "version": "0.4.0",
       "transport": {
         "type": "stdio"
       },
@@ -36,6 +36,13 @@
           "isRequired": false,
           "format": "string",
           "isSecret": false
+        },
+        {
+          "name": "GEMINI_INPUT_DIR",
+          "description": "Directory to resolve bare input-image filenames against",
+          "isRequired": false,
+          "format": "string",
+          "isSecret": false
         }
       ]
     }