sogni-gen 1.2.4 → 1.3.0

package/README.md

@@ -2,14 +2,45 @@
   <img src="screenshot.jpg" alt="Telegram image render workflow" width="320" />
 </p>
 
-# Image & Video gen for 🦞 OpenClaw
+# Sogni Gen — AI Image & Video Generation
 
 🎨 Generate **images and videos** using [Sogni AI](https://sogni.ai)'s decentralized GPU network.
 
-An [OpenClaw](https://github.com/OpenClaw/OpenClaw) plugin for AI image + video generation.
+Works as an [MCP server](https://modelcontextprotocol.io/) for **Claude Code** and **Claude Desktop**, and as an [OpenClaw](https://github.com/OpenClaw/OpenClaw) plugin.
 
 ## Installation
 
+### Claude Code (one command)
+
+```bash
+claude mcp add sogni -- npx -y -p sogni-gen sogni-gen-mcp
+```
+
+After install, just ask Claude things like:
+- "Generate an image of a sunset over mountains"
+- "Make a video of a cat playing piano"
+- "Edit this image to add a rainbow"
+- "Check my Sogni balance"
+
+### Claude Desktop
+
+Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:
+
+```json
+{
+  "mcpServers": {
+    "sogni": {
+      "command": "npx",
+      "args": ["-y", "-p", "sogni-gen", "sogni-gen-mcp"]
+    }
+  }
+}
+```
+
+Restart Claude Desktop after saving. The same natural-language commands work.
+
+> **Note:** Both Claude Code and Claude Desktop require Sogni credentials — see [Setup](#setup) below.
+
 ### Quick Install (OpenClaw) - Recommended
 
 Point your OpenClaw to the [`llm.txt`](https://raw.githubusercontent.com/Sogni-AI/openclaw-sogni-gen/main/llm.txt) and everything is set up — just paste the URL into Telegram, WhatsApp, or iMessages and the bot handles image and video generation automatically.
@@ -52,6 +83,7 @@ If OpenClaw loads this plugin, `sogni-gen` will read defaults from your OpenClaw
         "config": {
           "defaultImageModel": "z_image_turbo_bf16",
           "defaultEditModel": "qwen_image_edit_2511_fp8_lightning",
+          "defaultPhotoboothModel": "coreml-sogniXLturbo_alpha1_ad",
           "videoModels": {
             "t2v": "wan_v2.2-14b-fp8_t2v_lightx2v",
             "i2v": "wan_v2.2-14b-fp8_i2v_lightx2v",
@@ -123,6 +155,10 @@ node sogni-gen.mjs -m flux1-schnell-fp8 "a dragon eating tacos"
 # JPG output
 node sogni-gen.mjs --output-format jpg -o dragon.jpg "a dragon eating tacos"
 
+# Photobooth (face transfer)
+node sogni-gen.mjs --photobooth --ref face.jpg "80s fashion portrait"
+node sogni-gen.mjs --photobooth --ref face.jpg -n 4 "LinkedIn professional headshot"
+
 # Image edit with LoRA
 node sogni-gen.mjs -c subject.jpg --lora sogni_lora_v1 --lora-strength 0.7 \
   "add a neon cyberpunk glow"
@@ -161,6 +197,26 @@ node sogni-gen.mjs --video --estimate-video-cost --steps 20 \
   -m wan_v2.2-14b-fp8_t2v_lightx2v "ocean waves at sunset"
 ```
 
+## Photobooth (Face Transfer)
+
+Generate stylized portraits from a face photo using InstantID ControlNet:
+
+```bash
+# Basic photobooth
+node sogni-gen.mjs --photobooth --ref face.jpg "80s fashion portrait"
+
+# Multiple outputs
+node sogni-gen.mjs --photobooth --ref face.jpg -n 4 "LinkedIn professional headshot"
+
+# Custom ControlNet tuning
+node sogni-gen.mjs --photobooth --ref face.jpg --cn-strength 0.6 --cn-guidance-end 0.5 "oil painting"
+
+# Custom model
+node sogni-gen.mjs --photobooth --ref face.jpg -m coreml-dreamshaperXL_v21TurboDPMSDE "anime style"
+```
+
+Uses SDXL Turbo (`coreml-sogniXLturbo_alpha1_ad`) at 1024x1024 by default. The face image is passed via `--ref` and styled according to the prompt. Cannot be combined with `--video` or `-c/--context`.
+
 Multi-angle mode auto-builds the `<sks>` prompt and applies the `multiple_angles` LoRA.
 `--angles-360-video` generates i2v clips between consecutive angles (including last→first) and concatenates them with ffmpeg for a seamless loop.
 `--balance` / `--balances` does not require a prompt and exits after printing current `SPARK` and `SOGNI` balances.
@@ -221,7 +277,10 @@ Multi-angle mode auto-builds the `<sks>` prompt and applies the `multiple_angles
 --auto-resize-assets  Auto-resize video reference assets
 --no-auto-resize-assets  Disable auto-resize for video assets
 --estimate-video-cost Estimate video cost and exit (requires --steps)
---ref <path|url>      Reference image for i2v/s2v/animate
+--photobooth          Face transfer mode (InstantID + SDXL Turbo)
+--cn-strength <n>     ControlNet strength (default: 0.8)
+--cn-guidance-end <n> ControlNet guidance end point (default: 0.3)
+--ref <path|url>      Reference image for i2v/s2v/animate/photobooth
 --ref-end <path|url>  End frame for i2v interpolation
 --ref-audio <path>    Reference audio for s2v
 --ref-video <path>    Reference video for animate workflows
@@ -242,6 +301,7 @@ Multi-angle mode auto-builds the `<sks>` prompt and applies the `multiple_angles
 | `chroma-v.46-flash_fp8` | ~30s | Balanced |
 | `qwen_image_edit_2511_fp8` | ~30s | Image editing with context |
 | `qwen_image_edit_2511_fp8_lightning` | ~8s | Fast image editing |
+| `coreml-sogniXLturbo_alpha1_ad` | Fast | Photobooth face transfer (SDXL Turbo) |
 | `wan_v2.2-14b-fp8_t2v_lightx2v` | ~5min | Text-to-video |
 | `wan_v2.2-14b-fp8_i2v_lightx2v` | ~3-5min | Image-to-video |
 | `wan_v2.2-14b-fp8_s2v_lightx2v` | ~5min | Sound-to-video |

package/SKILL.md

@@ -110,7 +110,10 @@ node sogni-gen.mjs -q -o /tmp/cat.png "a cat wearing a hat"
 | `--auto-resize-assets` | Auto-resize video assets | true |
 | `--no-auto-resize-assets` | Disable auto-resize | - |
 | `--estimate-video-cost` | Estimate video cost and exit (requires --steps) | - |
-| `--ref <path>` | Reference image for video | required for video |
+| `--photobooth` | Face transfer mode (InstantID + SDXL Turbo) | - |
+| `--cn-strength <n>` | ControlNet strength (photobooth) | 0.8 |
+| `--cn-guidance-end <n>` | ControlNet guidance end point (photobooth) | 0.3 |
+| `--ref <path>` | Reference image for video or photobooth face | required for video/photobooth |
 | `--ref-end <path>` | End frame for i2v interpolation | - |
 | `--ref-audio <path>` | Reference audio for s2v | - |
 | `--ref-video <path>` | Reference video for animate workflows | - |
@@ -134,6 +137,7 @@ When installed as an OpenClaw plugin, `sogni-gen` will read defaults from:
         "config": {
           "defaultImageModel": "z_image_turbo_bf16",
           "defaultEditModel": "qwen_image_edit_2511_fp8_lightning",
+          "defaultPhotoboothModel": "coreml-sogniXLturbo_alpha1_ad",
           "videoModels": {
             "t2v": "wan_v2.2-14b-fp8_t2v_lightx2v",
             "i2v": "wan_v2.2-14b-fp8_i2v_lightx2v",
@@ -177,6 +181,7 @@ Seed strategies: `prompt-hash` (deterministic) or `random`.
 | `chroma-v.46-flash_fp8` | Medium | Balanced |
 | `qwen_image_edit_2511_fp8` | Medium | Image editing with context (up to 3) |
 | `qwen_image_edit_2511_fp8_lightning` | Fast | Quick image editing |
+| `coreml-sogniXLturbo_alpha1_ad` | Fast | Photobooth face transfer (SDXL Turbo) |
 
 ## Video Models
 
@@ -206,6 +211,32 @@ node sogni-gen.mjs --last-image "make it more vibrant"
 
 When context images are provided without `-m`, defaults to `qwen_image_edit_2511_fp8_lightning`.
 
+## Photobooth (Face Transfer)
+
+Generate stylized portraits from a face photo using InstantID ControlNet. When a user mentions "photobooth", wants a stylized portrait of themselves, or asks to transfer their face into a style, use `--photobooth` with `--ref` pointing to their face image.
+
+```bash
+# Basic photobooth
+node sogni-gen.mjs --photobooth --ref face.jpg "80s fashion portrait"
+
+# Multiple outputs
+node sogni-gen.mjs --photobooth --ref face.jpg -n 4 "LinkedIn professional headshot"
+
+# Custom ControlNet tuning
+node sogni-gen.mjs --photobooth --ref face.jpg --cn-strength 0.6 --cn-guidance-end 0.5 "oil painting"
+```
+
+Uses SDXL Turbo (`coreml-sogniXLturbo_alpha1_ad`) at 1024x1024 by default. The face image is passed via `--ref` and styled according to the prompt. Cannot be combined with `--video` or `-c/--context`.
+
+**Agent usage:**
+```bash
+# Photobooth: stylize a face photo
+node {{skillDir}}/sogni-gen.mjs -q --photobooth --ref /path/to/face.jpg -o /tmp/stylized.png "80s fashion portrait"
+
+# Multiple photobooth outputs
+node {{skillDir}}/sogni-gen.mjs -q --photobooth --ref /path/to/face.jpg -n 4 -o /tmp/stylized.png "LinkedIn professional headshot"
+```
+
 ## Multiple Angles (Turnaround)
 
 Generate specific camera angles from a single reference image using the Multiple Angles LoRA:
@@ -319,6 +350,11 @@ ls -la ~/.clawdbot/media/inbound/*.jpg | tail -3
 cp ~/.clawdbot/media/inbound/<latest>.jpg /tmp/to_restore.jpg
 ```
 
+## IMPORTANT KEYWORD RULE
+
+- If the user message includes the word "photobooth" (case-insensitive), always use `--photobooth` mode with `--ref` set to the user-provided face image.
+- Prioritize this rule over generic image-edit flows (`-c`) for that request.
+
 ## Agent Usage
 
 When user asks to generate/draw/create an image:
@@ -336,6 +372,9 @@ node {{skillDir}}/sogni-gen.mjs -q --video --ref /path/to/image.png -o /tmp/vide
 # Generate text-to-video
 node {{skillDir}}/sogni-gen.mjs -q --video -o /tmp/video.mp4 "ocean waves at sunset"
 
+# Photobooth: stylize a face photo
+node {{skillDir}}/sogni-gen.mjs -q --photobooth --ref /path/to/face.jpg -o /tmp/stylized.png "80s fashion portrait"
+
 # Check current SPARK/SOGNI balances (no prompt required)
 node {{skillDir}}/sogni-gen.mjs --json --balance
 

package/llm.txt

@@ -138,6 +138,7 @@ node {{skillDir}}/sogni-gen.mjs --json --balance
 
 ## Agent Behavior Guidelines
 
+0. If the user includes the keyword "photobooth" (case-insensitive), always use `--photobooth` with `--ref` to the user face image. Do not fall back to `-c` edit flow for that request.
 1. When the user asks to "draw", "generate", "create", or "make" an image: generate an image and send it.
 2. When they ask to "animate", "make a video", or "create a video": use --video mode.
 3. When they send a photo and ask to edit/change/modify it: use -c with their image.

package/mcp-server.mjs

@@ -0,0 +1,604 @@
+#!/usr/bin/env node
+/**
+ * sogni-gen MCP Server
+ *
+ * Exposes Sogni AI image/video generation as MCP tools for Claude Code
+ * and Claude Desktop.  Wraps the sogni-gen CLI using its --json mode.
+ *
+ * Install (Claude Code):
+ *   claude mcp add sogni -- npx -y -p sogni-gen sogni-gen-mcp
+ *
+ * Install (Claude Desktop – add to claude_desktop_config.json):
+ *   { "mcpServers": { "sogni": { "command": "npx", "args": ["-y", "-p", "sogni-gen", "sogni-gen-mcp"] } } }
+ */
+
+import { Server } from '@modelcontextprotocol/sdk/server/index.js';
+import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';
+import { CallToolRequestSchema, ListToolsRequestSchema } from '@modelcontextprotocol/sdk/types.js';
+import { spawn } from 'child_process';
+import { fileURLToPath } from 'url';
+import { dirname, join } from 'path';
+import { existsSync } from 'fs';
+import { homedir } from 'os';
+
+// ---------------------------------------------------------------------------
+// Paths
+// ---------------------------------------------------------------------------
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+const SOGNI_GEN = join(__dirname, 'sogni-gen.mjs');
+const CREDENTIALS_PATH = join(homedir(), '.config', 'sogni', 'credentials');
+
+// ---------------------------------------------------------------------------
+// CLI spawning helper
+// ---------------------------------------------------------------------------
+
+/**
+ * Spawn `node sogni-gen.mjs --json ...args`, collect stdout, parse JSON.
+ * Returns the parsed object on success or throws on failure.
+ */
+function runSogniGen(args, { timeoutMs = 30_000 } = {}) {
+  return new Promise((resolve, reject) => {
+    const child = spawn(process.execPath, [SOGNI_GEN, '--json', '--quiet', ...args], {
+      stdio: ['ignore', 'pipe', 'pipe'],
+      timeout: timeoutMs,
+    });
+
+    const stdoutChunks = [];
+    const stderrChunks = [];
+
+    child.stdout.on('data', (chunk) => stdoutChunks.push(chunk));
+    child.stderr.on('data', (chunk) => stderrChunks.push(chunk));
+
+    child.on('error', (err) => {
+      reject(new Error(`Failed to spawn sogni-gen: ${err.message}`));
+    });
+
+    child.on('close', (code) => {
+      const stdout = Buffer.concat(stdoutChunks).toString('utf8').trim();
+      const stderr = Buffer.concat(stderrChunks).toString('utf8').trim();
+
+      if (!stdout) {
+        reject(new Error(stderr || `sogni-gen exited with code ${code} and no output`));
+        return;
+      }
+
+      try {
+        const result = JSON.parse(stdout);
+        resolve(result);
+      } catch {
+        reject(new Error(`Failed to parse sogni-gen output: ${stdout.slice(0, 500)}`));
+      }
+    });
+  });
+}
+
+// ---------------------------------------------------------------------------
+// Credential check helper
+// ---------------------------------------------------------------------------
+
+function checkCredentials() {
+  if (existsSync(CREDENTIALS_PATH)) return null;
+  if (process.env.SOGNI_USERNAME && process.env.SOGNI_PASSWORD) return null;
+  return {
+    content: [
+      {
+        type: 'text',
+        text: [
+          'Sogni credentials not found. Please set up credentials:',
+          '',
+          '1. Create a Sogni account at https://sogni.ai',
+          '2. Create the credentials file:',
+          '',
+          '   mkdir -p ~/.config/sogni',
+          '   cat > ~/.config/sogni/credentials << \'EOF\'',
+          '   SOGNI_USERNAME=your_username',
+          '   SOGNI_PASSWORD=your_password',
+          '   EOF',
+          '   chmod 600 ~/.config/sogni/credentials',
+          '',
+          'Or set SOGNI_USERNAME and SOGNI_PASSWORD environment variables.',
+        ].join('\n'),
+      },
+    ],
+    isError: true,
+  };
+}
+
+// ---------------------------------------------------------------------------
+// Result formatting
+// ---------------------------------------------------------------------------
+
+function formatSuccess(result) {
+  const parts = [];
+
+  if (result.type === 'balance') {
+    parts.push(`SPARK: ${result.spark ?? 'N/A'}`);
+    parts.push(`SOGNI: ${result.sogni ?? 'N/A'}`);
+    return { content: [{ type: 'text', text: parts.join('\n') }] };
+  }
+
+  // Image / video result
+  if (result.prompt) parts.push(`Prompt: ${result.prompt}`);
+  parts.push(`Model: ${result.model}`);
+  parts.push(`Size: ${result.width}x${result.height}`);
+  if (result.seed != null) parts.push(`Seed: ${result.seed}`);
+
+  if (result.type === 'video') {
+    if (result.workflow) parts.push(`Workflow: ${result.workflow}`);
+    if (result.duration) parts.push(`Duration: ${result.duration}s`);
+    if (result.fps) parts.push(`FPS: ${result.fps}`);
+  }
+
+  if (result.localPath) parts.push(`Saved to: ${result.localPath}`);
+
+  // URLs
+  const urls = result.urls || [];
+  if (urls.length > 0) {
+    parts.push('');
+    urls.forEach((url, i) => {
+      parts.push(urls.length === 1 ? `URL: ${url}` : `URL #${i + 1}: ${url}`);
+    });
+  }
+
+  const content = [{ type: 'text', text: parts.join('\n') }];
+
+  // Also include image URLs as image content so Claude can see them
+  for (const url of urls) {
+    if (/\.(png|jpg|jpeg|webp|gif)(\?|$)/i.test(url)) {
+      content.push({ type: 'image', data: url, mimeType: 'image/png' });
+    }
+  }
+
+  return { content };
+}
+
+function formatError(result) {
+  const parts = [`Error: ${result.error}`];
+  if (result.errorCode) parts.push(`Code: ${result.errorCode}`);
+  if (result.hint) parts.push(`Hint: ${result.hint}`);
+  return { content: [{ type: 'text', text: parts.join('\n') }], isError: true };
+}
+
+function formatResult(result) {
+  if (result.success === false) return formatError(result);
+  return formatSuccess(result);
+}
+
+// ---------------------------------------------------------------------------
+// Tool definitions
+// ---------------------------------------------------------------------------
+
+const IMAGE_MODEL_TABLE = `Image Models:
+  z_image_turbo_bf16         — Fast (~5-10s), general purpose (default)
+  flux1-schnell-fp8          — Very fast (~3-5s), quick iterations
+  flux2_dev_fp8              — Slow (~2min), high quality
+  chroma-v.46-flash_fp8      — Medium (~30s), balanced
+  qwen_image_edit_2511_fp8   — Medium (~30s), image editing with context
+  qwen_image_edit_2511_fp8_lightning — Fast (~8s), quick image editing`;
+
+const VIDEO_MODEL_TABLE = `Video Models (auto-selected per workflow):
+  wan_v2.2-14b-fp8_t2v_lightx2v             — Text-to-video (~5min)
+  wan_v2.2-14b-fp8_i2v_lightx2v             — Image-to-video (~3-5min)
+  wan_v2.2-14b-fp8_s2v_lightx2v             — Sound-to-video (~5min)
+  wan_v2.2-14b-fp8_animate-move_lightx2v    — Animate-move (~5min)
+  wan_v2.2-14b-fp8_animate-replace_lightx2v — Animate-replace (~5min)`;
+
+const TOOLS = [
+  {
+    name: 'generate_image',
+    description: `Generate an image using Sogni AI's decentralized GPU network.
+
+${IMAGE_MODEL_TABLE}
+
+Cost: Uses Spark tokens. 512x512 is most cost-efficient. Claim 50 free daily Spark at https://app.sogni.ai/`,
+    inputSchema: {
+      type: 'object',
+      properties: {
+        prompt: {
+          type: 'string',
+          description: 'Image description / generation prompt',
+        },
+        model: {
+          type: 'string',
+          description: 'Model ID (default: z_image_turbo_bf16)',
+        },
+        width: {
+          type: 'number',
+          description: 'Image width in pixels (default: 512)',
+        },
+        height: {
+          type: 'number',
+          description: 'Image height in pixels (default: 512)',
+        },
+        count: {
+          type: 'number',
+          description: 'Number of images to generate (default: 1)',
+        },
+        seed: {
+          type: 'number',
+          description: 'Specific seed for reproducibility',
+        },
+        output: {
+          type: 'string',
+          description: 'Save image to this file path',
+        },
+        output_format: {
+          type: 'string',
+          enum: ['png', 'jpg'],
+          description: 'Output format (default: png)',
+        },
+        loras: {
+          type: 'array',
+          items: { type: 'string' },
+          description: 'LoRA model IDs',
+        },
+        lora_strengths: {
+          type: 'array',
+          items: { type: 'number' },
+          description: 'LoRA strengths (parallel to loras array)',
+        },
+      },
+      required: ['prompt'],
+    },
+  },
+  {
+    name: 'generate_video',
+    description: `Generate a video using Sogni AI's decentralized GPU network.
+
+Workflows:
+  t2v             — Text-to-video (default). Just provide a prompt.
+  i2v             — Image-to-video. Provide ref (reference image). Supports looping.
+  s2v             — Sound-to-video. Provide ref (face image) + ref_audio.
+  animate-move    — Transfer motion from ref_video to ref image.
+  animate-replace — Replace subject in ref_video with ref image.
+
+${VIDEO_MODEL_TABLE}
+
+Video dimensions must be divisible by 16, min 480px, max 1536px. Generation takes 3-5 minutes.
+Cost: Uses Spark tokens. Claim 50 free daily Spark at https://app.sogni.ai/`,
+    inputSchema: {
+      type: 'object',
+      properties: {
+        prompt: {
+          type: 'string',
+          description: 'Video description / generation prompt',
+        },
+        workflow: {
+          type: 'string',
+          enum: ['t2v', 'i2v', 's2v', 'animate-move', 'animate-replace'],
+          description: 'Video workflow (default: t2v, auto-inferred from provided refs)',
+        },
+        model: {
+          type: 'string',
+          description: 'Model ID (auto-selected per workflow by default)',
+        },
+        width: {
+          type: 'number',
+          description: 'Video width in pixels (default: 512, must be divisible by 16)',
+        },
+        height: {
+          type: 'number',
+          description: 'Video height in pixels (default: 512, must be divisible by 16)',
+        },
+        fps: {
+          type: 'number',
+          description: 'Frames per second (default: 16)',
+        },
+        duration: {
+          type: 'number',
+          description: 'Duration in seconds (default: 5)',
+        },
+        frames: {
+          type: 'number',
+          description: 'Override total frame count (alternative to duration)',
+        },
+        ref: {
+          type: 'string',
+          description: 'Reference image path or URL (for i2v, s2v, animate workflows)',
+        },
+        ref_end: {
+          type: 'string',
+          description: 'End frame image path or URL (for i2v interpolation)',
+        },
+        ref_audio: {
+          type: 'string',
+          description: 'Reference audio file path (for s2v workflow)',
+        },
+        ref_video: {
+          type: 'string',
+          description: 'Reference video file path (for animate workflows)',
+        },
+        seed: {
+          type: 'number',
+          description: 'Specific seed for reproducibility',
+        },
+        output: {
+          type: 'string',
+          description: 'Save video to this file path',
+        },
+        looping: {
+          type: 'boolean',
+          description: 'Generate seamless loop (i2v only)',
+        },
+      },
+      required: ['prompt'],
+    },
+  },
+  {
+    name: 'edit_image',
+    description: `Edit or transform an existing image using Sogni AI (Qwen image editing models).
+
+Provide 1-3 context images and a prompt describing the desired edit. Examples:
+  - "make the background a beach"
+  - "apply pop art style"
+  - "remove the person on the left"
+  - "add a rainbow in the sky"
+
+Models:
+  qwen_image_edit_2511_fp8_lightning — Fast (~8s), default
+  qwen_image_edit_2511_fp8          — Medium (~30s), higher quality`,
+    inputSchema: {
+      type: 'object',
+      properties: {
+        prompt: {
+          type: 'string',
+          description: 'Editing instruction describing the desired change',
+        },
+        context_images: {
+          type: 'array',
+          items: { type: 'string' },
+          description: 'Image file paths or URLs to edit (1-3 images)',
+          minItems: 1,
+          maxItems: 3,
+        },
+        model: {
+          type: 'string',
+          description: 'Model ID (default: qwen_image_edit_2511_fp8_lightning)',
+        },
+        width: {
+          type: 'number',
+          description: 'Output width in pixels',
+        },
+        height: {
+          type: 'number',
+          description: 'Output height in pixels',
+        },
+        output: {
+          type: 'string',
+          description: 'Save edited image to this file path',
+        },
+      },
+      required: ['prompt', 'context_images'],
+    },
+  },
+  {
+    name: 'photobooth',
+    description: `Generate stylized portraits from a face photo using InstantID face transfer.
+
+Provide a face reference image and a style prompt. Examples:
+  - "80s fashion portrait"
+  - "LinkedIn professional headshot"
+  - "oil painting Renaissance style"
+  - "anime character"
+
+Uses SDXL Turbo (coreml-sogniXLturbo_alpha1_ad) at 1024x1024 by default.
+The face likeness is preserved while applying the style from the prompt.`,
+    inputSchema: {
+      type: 'object',
+      properties: {
+        prompt: {
+          type: 'string',
+          description: 'Style/scene description for the portrait',
+        },
+        reference_face: {
+          type: 'string',
+          description: 'Face image file path or URL',
+        },
+        model: {
+          type: 'string',
+          description: 'Model ID (default: coreml-sogniXLturbo_alpha1_ad)',
+        },
+        cn_strength: {
+          type: 'number',
+          description: 'ControlNet strength — higher = more face likeness (default: 0.8)',
+        },
+        cn_guidance_end: {
+          type: 'number',
+          description: 'ControlNet guidance end point (default: 0.3)',
+        },
+        width: {
+          type: 'number',
+          description: 'Output width in pixels (default: 1024)',
+        },
+        height: {
+          type: 'number',
+          description: 'Output height in pixels (default: 1024)',
+        },
+        count: {
+          type: 'number',
+          description: 'Number of images to generate (default: 1)',
+        },
+        output: {
+          type: 'string',
+          description: 'Save image to this file path',
+        },
+      },
+      required: ['prompt', 'reference_face'],
+    },
+  },
+  {
+    name: 'check_balance',
+    description:
+      'Check your current Sogni token balances (SPARK and SOGNI). Free daily Spark tokens can be claimed at https://app.sogni.ai/',
+    inputSchema: {
+      type: 'object',
+      properties: {},
+    },
+  },
+  {
+    name: 'list_models',
+    description:
+      'List all available Sogni AI models for image generation, image editing, photobooth, and video generation with speed estimates.',
+    inputSchema: {
+      type: 'object',
+      properties: {},
+    },
+  },
+];
+
+// ---------------------------------------------------------------------------
+// Tool handlers
+// ---------------------------------------------------------------------------
+
+async function handleGenerateImage(params) {
+  const credErr = checkCredentials();
+  if (credErr) return credErr;
+
+  const args = [params.prompt];
+  if (params.model) args.push('-m', params.model);
+  if (params.width) args.push('-w', String(params.width));
+  if (params.height) args.push('-h', String(params.height));
+  if (params.count) args.push('-n', String(params.count));
+  if (params.seed != null) args.push('-s', String(params.seed));
+  if (params.output) args.push('-o', params.output);
+  if (params.output_format) args.push('--output-format', params.output_format);
+  if (params.loras?.length) args.push('--loras', params.loras.join(','));
+  if (params.lora_strengths?.length) args.push('--lora-strengths', params.lora_strengths.join(','));
+
+  const result = await runSogniGen(args, { timeoutMs: 60_000 });
+  return formatResult(result);
+}
+
+async function handleGenerateVideo(params) {
+  const credErr = checkCredentials();
+  if (credErr) return credErr;
+
+  const args = ['--video', params.prompt];
+  if (params.workflow) args.push('--workflow', params.workflow);
+  if (params.model) args.push('-m', params.model);
+  if (params.width) args.push('-w', String(params.width));
+  if (params.height) args.push('-h', String(params.height));
+  if (params.fps) args.push('--fps', String(params.fps));
+  if (params.duration) args.push('--duration', String(params.duration));
+  if (params.frames) args.push('--frames', String(params.frames));
+  if (params.ref) args.push('--ref', params.ref);
+  if (params.ref_end) args.push('--ref-end', params.ref_end);
+  if (params.ref_audio) args.push('--ref-audio', params.ref_audio);
+  if (params.ref_video) args.push('--ref-video', params.ref_video);
+  if (params.seed != null) args.push('-s', String(params.seed));
+  if (params.output) args.push('-o', params.output);
+  if (params.looping) args.push('--looping');
+
+  const result = await runSogniGen(args, { timeoutMs: 600_000 });
+  return formatResult(result);
+}
+
+async function handleEditImage(params) {
+  const credErr = checkCredentials();
+  if (credErr) return credErr;
+
+  const args = [];
+  for (const img of params.context_images) {
+    args.push('-c', img);
+  }
+  args.push(params.prompt);
+  if (params.model) args.push('-m', params.model);
+  if (params.width) args.push('-w', String(params.width));
+  if (params.height) args.push('-h', String(params.height));
+  if (params.output) args.push('-o', params.output);
+
+  const result = await runSogniGen(args, { timeoutMs: 60_000 });
+  return formatResult(result);
+}
+
+async function handlePhotobooth(params) {
+  const credErr = checkCredentials();
+  if (credErr) return credErr;
+
+  const args = ['--photobooth', '--ref', params.reference_face, params.prompt];
+  if (params.model) args.push('-m', params.model);
+  if (params.cn_strength != null) args.push('--cn-strength', String(params.cn_strength));
+  if (params.cn_guidance_end != null) args.push('--cn-guidance-end', String(params.cn_guidance_end));
+  if (params.width) args.push('-w', String(params.width));
+  if (params.height) args.push('-h', String(params.height));
+  if (params.count) args.push('-n', String(params.count));
+  if (params.output) args.push('-o', params.output);
+
+  const result = await runSogniGen(args, { timeoutMs: 60_000 });
+  return formatResult(result);
+}
+
+async function handleCheckBalance() {
+  const credErr = checkCredentials();
+  if (credErr) return credErr;
+
+  const result = await runSogniGen(['--balance'], { timeoutMs: 30_000 });
+  return formatResult(result);
+}
+
+function handleListModels() {
+  const text = `${IMAGE_MODEL_TABLE}
+
+Photobooth Model:
+  coreml-sogniXLturbo_alpha1_ad — Fast, face transfer (SDXL Turbo, default for --photobooth)
+
+${VIDEO_MODEL_TABLE}
+
+Defaults:
+  Image generation: z_image_turbo_bf16
+  Image editing:    qwen_image_edit_2511_fp8_lightning
+  Photobooth:       coreml-sogniXLturbo_alpha1_ad
+  Video:            auto-selected per workflow (t2v/i2v/s2v/animate-move/animate-replace)`;
+
+  return { content: [{ type: 'text', text }] };
+}
+
+// ---------------------------------------------------------------------------
+// Server setup
+// ---------------------------------------------------------------------------
+
+const server = new Server(
+  { name: 'sogni', version: '1.3.0' },
+  { capabilities: { tools: {} } },
+);
+
+server.setRequestHandler(ListToolsRequestSchema, async () => ({ tools: TOOLS }));
+
+server.setRequestHandler(CallToolRequestSchema, async (request) => {
+  const { name, arguments: params } = request.params;
+  try {
+    switch (name) {
+      case 'generate_image':
+        return await handleGenerateImage(params);
+      case 'generate_video':
+        return await handleGenerateVideo(params);
+      case 'edit_image':
+        return await handleEditImage(params);
+      case 'photobooth':
+        return await handlePhotobooth(params);
+      case 'check_balance':
+        return await handleCheckBalance();
+      case 'list_models':
+        return handleListModels();
+      default:
+        return {
+          content: [{ type: 'text', text: `Unknown tool: ${name}` }],
+          isError: true,
+        };
+    }
+  } catch (err) {
+    return {
+      content: [{ type: 'text', text: `Error: ${err.message}` }],
+      isError: true,
+    };
+  }
+});
+
+// ---------------------------------------------------------------------------
+// Start
+// ---------------------------------------------------------------------------
+
+const transport = new StdioServerTransport();
+await server.connect(transport);

package/package.json

@@ -1,11 +1,12 @@
 {
   "name": "sogni-gen",
-  "version": "1.2.4",
-  "description": "Sogni AI image generation plugin for OpenClaw",
+  "version": "1.3.0",
+  "description": "Sogni AI image & video generation — OpenClaw plugin and MCP server for Claude Code / Claude Desktop",
   "type": "module",
   "main": "sogni-gen.mjs",
   "bin": {
-    "sogni-gen": "sogni-gen.mjs"
+    "sogni-gen": "sogni-gen.mjs",
+    "sogni-gen-mcp": "mcp-server.mjs"
   },
   "scripts": {
     "test": "node --test test/*.test.mjs test/*.integration.mjs",
@@ -16,7 +17,11 @@
     "ai",
     "image-generation",
     "openclaw",
-    "plugin"
+    "plugin",
+    "mcp",
+    "claude",
+    "claude-code",
+    "claude-desktop"
   ],
   "author": "Mauvis Ledford",
   "license": "MIT",
@@ -24,6 +29,7 @@
     "node": ">=22.22.0"
   },
   "dependencies": {
+    "@modelcontextprotocol/sdk": "^1.0.0",
     "@sogni-ai/sogni-client-wrapper": "^1.4.2",
     "json5": "^2.2.3"
   },

package/sogni-gen.mjs

@@ -593,7 +593,10 @@ const options = {
   refAudio: null, // Reference audio for s2v
   refVideo: null, // Reference video for animate workflows
   contextImages: [], // Context images for image editing
-  looping: false // Create looping video (i2v only): generate A→B then B→A and concatenate
+  looping: false, // Create looping video (i2v only): generate A→B then B→A and concatenate
+  photobooth: false, // Photobooth mode (InstantID face transfer)
+  cnStrength: null, // ControlNet strength override
+  cnGuidanceEnd: null // ControlNet guidance end override
 };
 const cliSet = {
   output: false,
@@ -630,7 +633,10 @@ const cliSet = {
   refImageEnd: false,
   refAudio: false,
   refVideo: false,
-  context: false
+  context: false,
+  photobooth: false,
+  cnStrength: false,
+  cnGuidanceEnd: false
 };
 
 // Parse CLI args
@@ -762,6 +768,15 @@ for (let i = 0; i < args.length; i++) {
   } else if (arg === '-c' || arg === '--context') {
     options.contextImages.push(args[++i]);
     cliSet.context = true;
+  } else if (arg === '--photobooth') {
+    options.photobooth = true;
+    cliSet.photobooth = true;
+  } else if (arg === '--cn-strength') {
+    options.cnStrength = parseFloat(args[++i]);
+    cliSet.cnStrength = true;
+  } else if (arg === '--cn-guidance-end') {
+    options.cnGuidanceEnd = parseFloat(args[++i]);
+    cliSet.cnGuidanceEnd = true;
   } else if (arg === '--last-image') {
     // Use image from last render as reference/context
     if (existsSync(LAST_RENDER_PATH)) {
@@ -830,6 +845,12 @@ Image Options:
   -c, --context <path>  Context image for editing (can use multiple)
   --last-image          Use last generated image as context
 
+Photobooth (Face Transfer):
+  --photobooth            Face transfer mode (InstantID + SDXL Turbo)
+  --ref <path|url>        Face image (required with --photobooth)
+  --cn-strength <n>       ControlNet strength (default: 0.8)
+  --cn-guidance-end <n>   ControlNet guidance end point (default: 0.3)
+
 Video Options:
   --video, -v           Generate video instead of image
   --workflow <type>     Video workflow: t2v|i2v|s2v|animate-move|animate-replace
@@ -885,6 +906,8 @@ Examples:
   sogni-gen --video --last-image "gentle camera pan"
   sogni-gen -c photo.jpg "make the background a beach" -m qwen_image_edit_2511_fp8
   sogni-gen -c subject.jpg -c style.jpg "apply the style to the subject"
+  sogni-gen --photobooth --ref face.jpg "80s fashion portrait"
+  sogni-gen --photobooth --ref face.jpg -n 4 "LinkedIn professional headshot"
 `);
     process.exit(0);
   } else if (!arg.startsWith('-') && !options.prompt) {
@@ -1142,6 +1165,8 @@ if (options._lastImagePath) {
     } else if (!options.quiet) {
       console.error('Warning: --last-image ignored for text-to-video workflow.');
     }
+  } else if (options.photobooth) {
+    if (!options.refImage) options.refImage = options._lastImagePath;
   } else {
     options.contextImages.push(options._lastImagePath);
   }
@@ -1156,6 +1181,14 @@ if (options.video) {
   if (!cliSet.timeout && !timeoutFromConfig && options.timeout === 30000) {
     options.timeout = 300000; // 5 min for video
   }
+} else if (options.photobooth) {
+  // Photobooth uses SDXL Turbo + InstantID ControlNet
+  options.model = options.model || openclawConfig?.defaultPhotoboothModel || 'coreml-sogniXLturbo_alpha1_ad';
+  if (!cliSet.width) options.width = 1024;
+  if (!cliSet.height) options.height = 1024;
+  if (!cliSet.timeout && !timeoutFromConfig && options.timeout === 30000) {
+    options.timeout = 60000;
+  }
 } else if (options.contextImages.length > 0) {
   // Use qwen edit model when context images provided (unless model explicitly set)
   options.model = options.model || openclawConfig?.defaultEditModel || 'qwen_image_edit_2511_fp8_lightning';
@@ -1176,6 +1209,18 @@ if (!options.video && (options.refAudio || options.refVideo || options.videoWork
   });
 }
 
+if (options.photobooth) {
+  if (!options.refImage) {
+    fatalCliError('--photobooth requires --ref <face-image>.', { code: 'INVALID_ARGUMENT' });
+  }
+  if (options.video) {
+    fatalCliError('--photobooth cannot be combined with --video.', { code: 'INVALID_ARGUMENT' });
+  }
+  if (options.contextImages.length > 0) {
+    fatalCliError('--photobooth cannot be combined with -c/--context.', { code: 'INVALID_ARGUMENT' });
+  }
+}
+
 if (options.video) {
   if (options.videoWorkflow === 't2v') {
     if (options.refImage || options.refImageEnd || options.refAudio || options.refVideo) {
@@ -2416,6 +2461,53 @@ async function main() {
       }
       
       await client.createImageEditProject(editConfig);
+    } else if (options.photobooth) {
+      // Photobooth: face transfer with InstantID ControlNet
+      log(`Photobooth with ${options.model}...`);
+      if (options.seed !== null && options.seed !== undefined) log(`Using seed: ${options.seed}`);
+
+      const faceBuffer = await fetchMediaBuffer(options.refImage);
+      const modelDefaults = getModelDefaults(options.model, openclawConfig);
+      const steps = options.steps ?? modelDefaults?.steps ?? 7;
+      const guidance = options.guidance ?? modelDefaults?.guidance ?? 2;
+
+      const projectConfig = {
+        modelId: options.model,
+        positivePrompt: options.prompt,
+        negativePrompt: '',
+        stylePrompt: '',
+        numberOfMedia: options.count,
+        tokenType: options.tokenType || 'spark',
+        waitForCompletion: false,
+        sizePreset: 'custom',
+        width: options.width,
+        height: options.height,
+        steps,
+        guidance,
+        disableNSFWFilter: true,
+        sampler: options.sampler || 'dpmpp_sde',
+        scheduler: options.scheduler || 'karras',
+        controlNet: {
+          name: 'instantid',
+          image: faceBuffer,
+          strength: options.cnStrength ?? 0.7,
+          mode: 'balanced',
+          guidanceStart: 0,
+          guidanceEnd: options.cnGuidanceEnd ?? 0.6,
+        }
+      };
+
+      if (options.outputFormat) projectConfig.outputFormat = options.outputFormat;
+      if (options.seed !== null && options.seed !== undefined) projectConfig.seed = options.seed;
+      if (options.loras.length > 0) projectConfig.loras = options.loras;
+      if (options.loraStrengths.length > 0) projectConfig.loraStrengths = options.loraStrengths;
+
+      const projectResult = await client.createImageProject(projectConfig);
+
+      // Check for errors in the response (e.g., insufficient tokens)
+      if (projectResult?.error || projectResult?.message) {
+        throw new Error(projectResult.error || projectResult.message);
+      }
     } else {
       // Standard image generation
       log(`Generating with ${options.model}...`);
@@ -2513,6 +2605,10 @@ async function main() {
       if (options.contextImages.length > 0) {
         renderInfo.contextImages = options.contextImages;
       }
+      if (options.photobooth) {
+        renderInfo.photobooth = true;
+        renderInfo.refImage = options.refImage;
+      }
       saveLastRender(renderInfo);
       
       // Save to file if requested
@@ -2700,6 +2796,15 @@ async function main() {
         if (options.contextImages.length > 0) {
           output.contextImages = options.contextImages;
         }
+        if (options.photobooth) {
+          output.photobooth = true;
+          output.refImage = options.refImage;
+          output.controlNet = {
+            name: 'instantid',
+            strength: options.cnStrength ?? 0.7,
+            guidanceEnd: options.cnGuidanceEnd ?? 0.6,
+          };
+        }
         console.log(JSON.stringify(output));
       } else {
         urls.forEach(url => console.log(url));