internal/

Package audio provides model registry and local inference for audio processing (TTS and ASR) using ONNX models with pure-onnx runtime.

Package background 实现基于 RMBG 2.0 语义分割的图片去背与替换。

Package client implements API client for Midjourney API endpoints.

Package config handles loading and merging of YAML configuration.

Package forensic provides multi-signal fusion for AI-generated content detection.

Package ideas provides BM25-powered search over a local prompt ideas dataset.

Package mcp implements an MCP (Model Context Protocol) server.

Package ocr provides offline text detection and recognition using ONNX Runtime.

Package onnx provides a pure-Go wrapper around the ONNX Runtime shared library for AIGC image detection.

Package onnxrt handles ONNX Runtime shared library download, extraction, and path resolution.

Package pdf provides PDF processing in Go: text extraction for born-digital PDFs via ledongthuc/pdf (pure Go), and page-to-image rendering via mutool (external CLI, muPDF tools — required only for scanned/image-based PDFs).

Package provider detects which AI provider the user has configured based on the API base URL.

Package rmbg 提供基于 RMBG 2.0 (Remove Background 2.0) ONNX 模型的语义分割去背功能。

Package service provides shared business logic used by CLI commands.

Package types defines request/response data structures for the Midjourney API.

Package vision provides local image understanding via ONNX-based VLM models (Florence-2).

Package watermark provides visible AI watermark detection and removal.