Identify GPU adapters, driver versions, Vulkan SDK presence, shader compilers, and Vulkan API version to prevent rendering code from guessing unsupported features.
List all directories currently accessible to the server, enabling you to register new directories and understand the filesystem scope before performing file operations.
Provides offline, privacy-preserving image recognition, OCR, and scene description for AI assistants via the Model Context Protocol, with Vulkan-accelerated local processing.
Enables AI coding agents to test and automate Android apps rendering UI in OpenGL/Vulkan/Metal surfaces via screenshot-based vision tools for tapping, swiping, long-pressing, and recording, without accessibility tree dependency.
Read text and document files (PDF, DOCX, PPTX, XLSX, ODT, ODP, ODS) with flexible modes: full file, first N lines, last N lines, or a specific line range. Extracts PDF content with format metadata.
Rebuild the documentation embedding index for semantic search. Use clean mode to rebuild from scratch or index only missing docs, with optional slug selection.
Compiles shader files to SPIR-V or DXIL with automatic compiler selection by extension. Reports errors with file and line, supporting Vulkan and DirectX targets.
Batch read multiple files concurrently with per-file mode control. Each file can specify full read, first N lines, last N lines, or an arbitrary line range. Supports text and documents like PDF, DOCX.
Edit text and code files with intelligent matching strategies. Supports single-file or batch multi-file edits with dry-run preview and atomic rollback.
Find files and directories recursively using glob patterns. Returns absolute paths to all matched items, solving the problem of locating files when exact paths are unknown.
Upscale an image to higher resolution using real-ESRGAN with Vulkan acceleration. Requires a discrete GPU for practical performance; integrated graphics may be slow.
Enlarge and enhance local images using Real-ESRGAN with GPU acceleration, no login required. Outputs a high-resolution PNG at up to 4x scale, with model options for general photos or anime.
Write multiple files of various types (text, PDF, DOCX) concurrently. Auto-detects file type by extension and supports HTML formatting for PDF/DOCX. Failed writes for individual files don't halt the batch.