{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"github-mochidiffusion--mochidiffusion","slug":"mochidiffusion--mochidiffusion","name":"MochiDiffusion","type":"repo","url":"https://github.com/MochiDiffusion/MochiDiffusion","page_url":"https://unfragile.ai/mochidiffusion--mochidiffusion","categories":["image-generation"],"tags":["ane","apple","apple-silicon","coreml","macos","neural-engine","stable-diffusion","swift","swiftui"],"pricing":{"model":"open_source","free":true,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"github-mochidiffusion--mochidiffusion__cap_0","uri":"capability://image.visual.neural.engine.optimized.stable.diffusion.inference","name":"neural engine-optimized stable diffusion inference","description":"Executes Stable Diffusion image generation models directly on Apple Silicon's Neural Engine using Core ML framework, leveraging split_einsum model optimization to distribute computation across CPU, GPU, and Neural Engine. The pipeline chains multiple Core ML models (text encoder, UNet denoiser, VAE decoder) with custom scheduling logic to minimize memory footprint (~150MB) while maximizing throughput through hardware-specific compute unit selection.","intents":["Run Stable Diffusion locally without cloud dependencies or API costs","Generate images with minimal memory overhead on MacBook Air/Pro","Achieve fast inference by utilizing Apple Silicon's specialized neural hardware"],"best_for":["macOS developers building offline image generation workflows","Mac users prioritizing privacy and latency over cloud-based generation","Teams deploying on-device ML without internet connectivity requirements"],"limitations":["Requires Core ML model conversion from PyTorch/ONNX format — not all Stable Diffusion variants are pre-converted","split_einsum optimization adds model size overhead (~2-3x larger than original) but enables Neural Engine execution","Limited to Apple Silicon Macs (M1/M2/M3+) — Intel Macs fall back to CPU-only inference with significant performance degradation","No support for arbitrary custom LoRA/embedding injection at runtime — models must be pre-baked into Core ML format"],"requires":["macOS 12.0+ with Apple Silicon (M1, M2, M3, M4 chips)","Minimum 8GB RAM (16GB+ recommended for batch generation)","Core ML-formatted Stable Diffusion models (v1.5, v2.1, SDXL variants available in repo)"],"input_types":["text prompts (UTF-8 strings up to model tokenizer limit)","negative prompts (optional, same format)","seed (integer for reproducibility)","guidance scale (float 1.0-20.0)","number of inference steps (integer 20-50 typical)"],"output_types":["PNG images with EXIF metadata containing prompt/parameters","Raw pixel data (RGBA format)","Generation metadata (seed, steps, timing)"],"categories":["image-visual","hardware-acceleration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_1","uri":"capability://image.visual.image.to.image.generation.with.reference.guidance","name":"image-to-image generation with reference guidance","description":"Accepts an existing image as input and generates variations by injecting the reference image's latent representation into the diffusion process at a configurable noise level (strength parameter). The VAE encoder converts the input image to latent space, the UNet denoiser applies conditional diffusion starting from the noisy latent, and the VAE decoder reconstructs the final image. Strength parameter (0.0-1.0) controls how much the output diverges from the input: low values preserve composition, high values enable radical transformation.","intents":["Create variations of existing images while preserving composition or style","Perform style transfer by providing a reference image with text prompt","Iteratively refine generated images through multiple passes"],"best_for":["Creative professionals iterating on image concepts","Developers building image editing workflows with AI enhancement","Users performing style transfer without external tools"],"limitations":["Strength parameter is global — cannot selectively preserve regions (no inpainting mask support in base implementation)","Input image must be resized to model's native resolution (512x512 or 768x768) — aspect ratio changes may distort composition","VAE encoding/decoding adds ~500ms latency per image on top of diffusion time","Quality degrades if input image is significantly different from training data distribution"],"requires":["Input image in PNG, JPEG, or HEIC format","Image dimensions compatible with model (typically 512x512 or 768x768)","Text prompt describing desired output","Strength parameter (0.0-1.0, default 0.7)"],"input_types":["image file (PNG, JPEG, HEIC)","text prompt","strength (float 0.0-1.0)","seed (optional, for reproducibility)"],"output_types":["PNG image with EXIF metadata","Generation parameters (strength, seed, prompt)"],"categories":["image-visual","image-editing"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_10","uri":"capability://text.generation.language.internationalization.and.multi.language.ui.support","name":"internationalization and multi-language ui support","description":"Implements localization for UI strings, help text, and documentation in multiple languages (English, Chinese, Korean, etc.) using Xcode's localization system (.strings files and Localizable.strings). Language selection is automatic based on system locale but can be overridden in settings. All UI elements (buttons, labels, prompts) are localized; documentation is provided in multiple languages via README files.","intents":["Support non-English users with native-language UI","Provide localized documentation and help text","Enable language selection in app settings"],"best_for":["International teams deploying to multiple regions","Open-source projects supporting global communities","Developers building multi-language macOS apps"],"limitations":["Localization maintenance burden increases with language count — each new language requires translation review","Some technical terms may not have direct translations — fallback to English may be necessary","Localization files can become out of sync with source code if not carefully managed","RTL languages (Arabic, Hebrew) require additional UI layout adjustments not fully implemented"],"requires":["Localizable.strings files for each language","Language code (en, zh-Hans, ko, etc.)","System locale or user preference"],"input_types":["language code (string)","UI string key (string)"],"output_types":["localized UI string","localized documentation"],"categories":["text-generation-language","localization"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_11","uri":"capability://automation.workflow.sparkle.based.automatic.update.system.with.version.checking","name":"sparkle-based automatic update system with version checking","description":"Integrates Sparkle framework for automatic app updates, checking for new versions on app launch and periodically in background. Updates are downloaded silently and installed on next app restart with user notification. Update manifest (appcast.xml) is hosted on GitHub and specifies available versions, download URLs, and release notes. Users can manually check for updates or disable automatic checking in settings.","intents":["Deliver bug fixes and feature updates to users automatically","Notify users of new versions without requiring manual checking","Enable rollback or skip of specific versions if needed"],"best_for":["Open-source projects distributing via GitHub","Teams deploying macOS apps with frequent updates","Users wanting automatic security and feature updates"],"limitations":["Sparkle requires internet connectivity to check for updates — offline users won't receive notifications","Update manifest must be manually maintained and hosted — no automatic CI/CD integration in base Sparkle","Users cannot selectively update features — all-or-nothing update model","Sparkle is macOS-only — no iOS/iPadOS support"],"requires":["Sparkle framework (open-source, bundled in app)","appcast.xml manifest hosted on web server (GitHub, etc.)","Signed app bundle (code signing required for security)"],"input_types":["version number (string)","download URL (string)","release notes (HTML or text)"],"output_types":["update notification (user-facing alert)","downloaded app bundle","installation status"],"categories":["automation-workflow","software-distribution"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_12","uri":"capability://memory.knowledge.custom.model.import.and.directory.based.model.discovery","name":"custom model import and directory-based model discovery","description":"Enables users to import custom Core ML Stable Diffusion models from local directories without recompiling the app. The system scans a designated models directory (in app bundle or user Documents) for .mlmodel or .mlpackage files, automatically detects model type (split_einsum vs. original) and architecture (v1.5, v2.1, SDXL), and makes them available in the model selection UI. Model metadata (name, size, compute unit compatibility) is extracted from file attributes and model bundle info.","intents":["Use custom fine-tuned or community models without app recompilation","Experiment with different model architectures and variants","Share models between users via directory sync or cloud storage"],"best_for":["Researchers experimenting with custom model variants","Teams sharing fine-tuned models across users","Developers building extensible image generation apps"],"limitations":["Model format must be Core ML (.mlmodel or .mlpackage) — PyTorch/ONNX models require external conversion","Model discovery is filesystem-based — no validation of model correctness or compatibility","Large model files (2-7GB) can be slow to copy to models directory","No versioning or dependency management — users must manually manage model variants"],"requires":["Core ML model files (.mlmodel or .mlpackage)","Models directory (app bundle or user Documents/MochiDiffusion/Models)","Model metadata (name, type, architecture) inferred from filename or bundle info"],"input_types":["model file path (local filesystem)","model metadata (optional, inferred from file)"],"output_types":["model list (available models in UI)","model metadata (name, size, compute unit compatibility)","model selection (user choice)"],"categories":["memory-knowledge","model-management"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_2","uri":"capability://image.visual.controlnet.guided.generation.with.structural.conditioning","name":"controlnet-guided generation with structural conditioning","description":"Integrates ControlNet models (separate Core ML networks) into the diffusion pipeline to provide structural guidance via edge maps, depth maps, pose skeletons, or other conditioning inputs. The ControlNet processes the conditioning image in parallel with the main UNet, producing cross-attention guidance that steers generation toward matching the structural constraints. Multiple ControlNet models can be loaded and weighted independently, enabling composition of multiple constraints (e.g., pose + depth).","intents":["Generate images matching specific structural layouts (pose, depth, edges)","Maintain consistent composition across multiple generations","Enforce spatial constraints without manual masking"],"best_for":["Game developers generating character poses matching motion capture data","Architects visualizing designs with depth/perspective constraints","Content creators maintaining consistent spatial composition across batches"],"limitations":["ControlNet models must be pre-converted to Core ML format — limited availability compared to PyTorch ecosystem","Conditioning image preprocessing (edge detection, depth estimation) must be done externally or via bundled utilities","Multiple ControlNets increase memory footprint and latency proportionally (each adds ~50-100MB and ~200ms per step)","Quality depends heavily on conditioning image quality — noisy/incorrect conditioning produces artifacts"],"requires":["Core ML-formatted ControlNet model (edge, depth, pose, or other variant)","Conditioning image (same resolution as base model, typically 512x512)","Text prompt","ControlNet weight (0.0-1.0, controls influence strength)"],"input_types":["conditioning image (PNG, JPEG, HEIC)","controlnet model identifier","text prompt","controlnet weight (float 0.0-1.0)"],"output_types":["PNG image with EXIF metadata","Generation parameters including ControlNet weight and model"],"categories":["image-visual","conditional-generation"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_3","uri":"capability://image.visual.real.esrgan.upscaling.with.neural.super.resolution","name":"real-esrgan upscaling with neural super-resolution","description":"Applies Real-ESRGAN neural network model (converted to Core ML) to generated or imported images to increase resolution by 2x or 4x while enhancing detail and reducing artifacts. The upscaler processes images in tiles to manage memory constraints, applies learned super-resolution kernels, and blends tile boundaries to avoid seams. Upscaling runs asynchronously in the job queue to avoid blocking UI.","intents":["Increase resolution of 512x512 generated images to 1024x1024 or 2048x2048","Enhance detail and reduce compression artifacts in existing images","Prepare generated images for print or high-resolution display"],"best_for":["Users generating images for print or large-format display","Workflows requiring high-resolution output from lower-resolution models","Batch processing pipelines upscaling multiple images"],"limitations":["Upscaling adds 2-5 seconds per image (2x) or 5-10 seconds (4x) depending on input size","Tile-based processing may introduce subtle seams at tile boundaries despite blending","Cannot recover information not present in original image — upscaling is detail enhancement, not hallucination","Memory usage scales with output resolution — 4x upscaling of 512x512 image requires ~500MB temporary memory"],"requires":["Core ML Real-ESRGAN model (2x or 4x variant)","Input image (any resolution, typically 512x512 or larger)","Upscaling factor (2x or 4x)"],"input_types":["image file (PNG, JPEG, HEIC)","upscaling factor (2 or 4)"],"output_types":["PNG image at 2x or 4x original resolution","Upscaling metadata (factor, processing time)"],"categories":["image-visual","image-enhancement"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_4","uri":"capability://automation.workflow.asynchronous.job.queue.with.progress.tracking.and.cancellation","name":"asynchronous job queue with progress tracking and cancellation","description":"Manages sequential or parallel image generation tasks in a queue system, tracking progress per job (step count, ETA, memory usage) and enabling cancellation mid-generation. Jobs are persisted to disk and survive app restart. The queue system decouples UI from long-running inference, allowing users to queue multiple generations and interact with the app while processing occurs. Progress updates stream to UI via SwiftUI state bindings.","intents":["Queue multiple image generations without blocking UI","Monitor generation progress with step-by-step updates and ETA","Cancel in-progress generations and restart from queue","Persist job queue across app restarts"],"best_for":["Users batch-generating multiple images overnight or during work","Workflows requiring reliable job persistence and recovery","Developers building image generation apps with progress UX"],"limitations":["Queue is in-memory with disk persistence — large queues (100+ jobs) may consume significant RAM","Cancellation stops current job but doesn't interrupt mid-step inference (waits for current step to complete, ~1-2 second delay)","No priority queue or job reordering — jobs execute in FIFO order","Progress ETA is estimated based on average step time — actual time varies with model and hardware state"],"requires":["Job queue data structure (stored in app's Documents directory)","Generation parameters per job (prompt, seed, model, etc.)","Async/await or callback-based progress reporting"],"input_types":["generation job (prompt, model, parameters)","queue operation (enqueue, dequeue, cancel)"],"output_types":["progress updates (current step, total steps, ETA, memory usage)","completed image with metadata","job status (queued, running, completed, cancelled)"],"categories":["automation-workflow","task-management"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_5","uri":"capability://image.visual.exif.metadata.preservation.and.embedding.in.generated.images","name":"exif metadata preservation and embedding in generated images","description":"Automatically embeds generation parameters (prompt, negative prompt, seed, model name, guidance scale, steps, ControlNet settings) into PNG/JPEG EXIF metadata when saving images. Metadata is human-readable and machine-parseable, enabling downstream tools to reproduce generations or extract parameters for analysis. Metadata is preserved when images are exported or shared.","intents":["Record generation parameters for reproducibility and auditing","Enable downstream tools to extract and reuse generation settings","Share images with embedded context about how they were created"],"best_for":["Researchers tracking generation parameters for reproducibility studies","Content creators documenting their generation workflows","Teams sharing generated images with embedded context"],"limitations":["EXIF metadata is stripped by some image processors and social media platforms","Metadata size is limited by EXIF spec (~64KB) — very long prompts may be truncated","PNG EXIF support is less universal than JPEG — some tools may not read PNG metadata","No encryption or signing of metadata — parameters can be modified by any tool"],"requires":["Image file format supporting EXIF (PNG, JPEG, HEIC)","Generation parameters (prompt, seed, model, etc.)","EXIF writing library (built into Core Image/ImageIO)"],"input_types":["generated image (PNG, JPEG, HEIC)","generation parameters (dict/struct)"],"output_types":["image file with embedded EXIF metadata","metadata dict (extractable by other tools)"],"categories":["image-visual","metadata-management"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_6","uri":"capability://memory.knowledge.core.ml.model.management.with.compute.unit.selection","name":"core ml model management with compute unit selection","description":"Manages loading, caching, and selection of Core ML Stable Diffusion models with automatic compute unit assignment (CPU, GPU, Neural Engine). The system detects model type (split_einsum vs. original) and selects optimal compute unit based on model architecture and available hardware. Models are lazy-loaded on first use and cached in memory to avoid repeated disk I/O. Custom models can be imported from user-specified directories.","intents":["Load and switch between multiple Stable Diffusion model variants","Automatically optimize inference by selecting appropriate compute unit","Import custom Core ML models without recompiling app","Cache models in memory to minimize load time between generations"],"best_for":["Users experimenting with different model architectures (v1.5, v2.1, SDXL)","Developers building model management UIs","Teams deploying custom fine-tuned models"],"limitations":["Model files are large (2-4GB for Stable Diffusion v1.5, 5-7GB for SDXL) — storage and loading time are significant","Compute unit selection is automatic and not user-configurable — no override for testing","Only Core ML format supported — PyTorch/ONNX models require external conversion","Model caching is in-memory only — switching models requires full reload (30-60 seconds)"],"requires":["Core ML model files (.mlmodel or .mlpackage format)","Model metadata (name, type, supported compute units)","Models directory in app bundle or user Documents folder"],"input_types":["model file path (local or bundled)","model metadata (name, type, compute unit preference)"],"output_types":["loaded Core ML model (MLModel instance)","compute unit assignment (CPU, GPU, Neural Engine)","model metadata (name, size, load time)"],"categories":["memory-knowledge","model-management"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_7","uri":"capability://planning.reasoning.scheduler.based.diffusion.step.control","name":"scheduler-based diffusion step control","description":"Implements multiple noise scheduling algorithms (DDPM, DDIM, Euler, Karras) that control the diffusion process across inference steps. The scheduler determines noise levels at each step, enabling trade-offs between quality and speed. Users can select scheduler and number of steps (typically 20-50); fewer steps reduce latency but may reduce quality. Scheduler is applied uniformly across all generation modes (text-to-image, image-to-image, ControlNet).","intents":["Trade off generation quality vs. speed by adjusting step count","Experiment with different noise schedules for quality tuning","Achieve consistent results across different generation modes"],"best_for":["Users optimizing generation speed for real-time workflows","Researchers comparing scheduler effects on output quality","Developers tuning quality/latency tradeoffs"],"limitations":["Scheduler selection is global — cannot vary per-generation","Fewer steps (20) produce faster but lower-quality results; more steps (50+) improve quality but add latency","Some schedulers (Karras) require specific step counts for optimal results","Scheduler changes require model recompilation in some cases (not true for Core ML inference)"],"requires":["Scheduler type (DDPM, DDIM, Euler, Karras, etc.)","Number of steps (integer 20-50 typical, up to 100)","Guidance scale (float 1.0-20.0)"],"input_types":["scheduler name (string)","step count (integer)","guidance scale (float)"],"output_types":["noise schedule (array of noise levels per step)","generation parameters (scheduler, steps, guidance)"],"categories":["planning-reasoning","inference-control"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_8","uri":"capability://automation.workflow.swiftui.based.native.macos.ui.with.gallery.and.sidebar.controls","name":"swiftui-based native macos ui with gallery and sidebar controls","description":"Implements native macOS user interface using SwiftUI framework with three main sections: gallery view (grid of generated images with metadata), sidebar controls (prompt input, model selection, generation parameters), and inspector panel (detailed image metadata and export options). UI is responsive to generation progress via SwiftUI state bindings, updating in real-time as jobs complete. Sidebar controls are context-aware, showing relevant options based on selected generation mode (text-to-image, image-to-image, ControlNet).","intents":["Provide native macOS experience with familiar UI patterns","Enable rapid iteration with quick access to generation parameters","Visualize generation history and metadata in organized gallery","Export and manage generated images with metadata"],"best_for":["macOS users expecting native app experience","Developers building SwiftUI-based image generation UIs","Teams deploying on macOS with consistent design language"],"limitations":["SwiftUI has performance limitations with large galleries (100+ images) — scrolling may stutter","State management complexity increases with feature count — large apps may require refactoring to MVVM","SwiftUI debugging tools are less mature than UIKit/AppKit — troubleshooting can be difficult","macOS-only — no iOS/iPadOS support without significant refactoring"],"requires":["macOS 12.0+ (SwiftUI 3.0+)","Xcode 14.0+ for development","SwiftUI state management (ObservedObject, StateObject, etc.)"],"input_types":["user input (text prompts, parameter sliders)","image selection (gallery tap, drag-drop)","menu actions (export, delete, duplicate)"],"output_types":["rendered UI (SwiftUI views)","user actions (generation request, image selection)","state updates (progress, completion)"],"categories":["automation-workflow","user-interface"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"github-mochidiffusion--mochidiffusion__cap_9","uri":"capability://data.processing.analysis.image.storage.and.gallery.management.with.local.persistence","name":"image storage and gallery management with local persistence","description":"Manages persistent storage of generated images in app's Documents directory with SQLite or plist-based metadata index. Gallery view loads images lazily from disk, caching thumbnails in memory for fast scrolling. Images are organized by generation date and searchable by prompt text. Deletion removes both image file and metadata entry. Export functionality copies images to user-selected locations with metadata preservation.","intents":["Organize and browse generated images with metadata","Search generation history by prompt or parameters","Export images with embedded metadata to external locations","Manage disk space by deleting unwanted generations"],"best_for":["Users maintaining large generation libraries (100+ images)","Workflows requiring searchable generation history","Teams sharing generated images with embedded context"],"limitations":["Local storage only — no cloud sync or backup (requires manual export)","Search is text-based on prompts only — no content-based image search","Large galleries (1000+ images) may have slow load times due to metadata parsing","Thumbnail caching uses memory — very large galleries may consume significant RAM"],"requires":["App Documents directory with write permissions","Metadata storage (SQLite or plist)","Image files (PNG, JPEG, HEIC)"],"input_types":["generated image file","generation metadata (prompt, seed, model, etc.)","search query (text)"],"output_types":["image file (with or without metadata)","metadata dict","search results (filtered image list)"],"categories":["data-processing-analysis","file-management"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":46,"verified":false,"data_access_risk":"high","permissions":["macOS 12.0+ with Apple Silicon (M1, M2, M3, M4 chips)","Minimum 8GB RAM (16GB+ recommended for batch generation)","Core ML-formatted Stable Diffusion models (v1.5, v2.1, SDXL variants available in repo)","Input image in PNG, JPEG, or HEIC format","Image dimensions compatible with model (typically 512x512 or 768x768)","Text prompt describing desired output","Strength parameter (0.0-1.0, default 0.7)","Localizable.strings files for each language","Language code (en, zh-Hans, ko, etc.)","System locale or user preference"],"failure_modes":["Requires Core ML model conversion from PyTorch/ONNX format — not all Stable Diffusion variants are pre-converted","split_einsum optimization adds model size overhead (~2-3x larger than original) but enables Neural Engine execution","Limited to Apple Silicon Macs (M1/M2/M3+) — Intel Macs fall back to CPU-only inference with significant performance degradation","No support for arbitrary custom LoRA/embedding injection at runtime — models must be pre-baked into Core ML format","Strength parameter is global — cannot selectively preserve regions (no inpainting mask support in base implementation)","Input image must be resized to model's native resolution (512x512 or 768x768) — aspect ratio changes may distort composition","VAE encoding/decoding adds ~500ms latency per image on top of diffusion time","Quality degrades if input image is significantly different from training data distribution","Localization maintenance burden increases with language count — each new language requires translation review","Some technical terms may not have direct translations — fallback to English may be necessary","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.6111820134079844,"quality":0.35,"ecosystem":0.6000000000000001,"match_graph":0.25,"freshness":0.75,"weights":{"adoption":0.3,"quality":0.2,"ecosystem":0.15,"match_graph":0.3,"freshness":0.05}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-05-24T12:16:22.062Z","last_scraped_at":"2026-05-03T13:58:42.318Z","last_commit":"2026-04-26T00:12:17Z"},"community":{"stars":7892,"forks":364,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=mochidiffusion--mochidiffusion","compare_url":"https://unfragile.ai/compare?artifact=mochidiffusion--mochidiffusion"}},"signature":"rhdkaG7gYeNpwvf6M1jlSl3DP8ODRQ4dIEztgWpL+mRHfMBNOUbvtHRNgkqQJqr21DyxeWVX3YyhOTl+9atWCw==","signedAt":"2026-06-21T13:00:22.077Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/mochidiffusion--mochidiffusion","artifact":"https://unfragile.ai/mochidiffusion--mochidiffusion","verify":"https://unfragile.ai/api/v1/verify?slug=mochidiffusion--mochidiffusion","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}