AntheaLaffy

Convert Markdown with LaTeX math, images, and tables into a polished PDF. Two rendering engines: (1) weasyprint — pure Python, no browser needed; (2) chromium — best for matrices/bmatrix/pmatrix/vmatrix and complex formulas. Supports Chinese/Japanese fonts. Can chain after ocr-md-polish for a complete OCR→clean→PDF workflow.

Burn core framework - Provides tensor operations, automatic differentiation, and neural network building blocks for Rust deep learning.

Burn CUDA backend - High-performance NVIDIA GPU acceleration. Provides optimal performance for NVIDIA hardware.

Burn Metal backend - Apple GPU acceleration for macOS and iOS. Optimized for Apple Silicon and Intel-based Macs.

Burn NdArray backend - CPU-based tensor operations using Rust's ndarray crate. Ideal for development, testing, and lightweight inference.

Burn ONNX support - Import ONNX models into Burn for inference on any backend. Supports compile-time code generation and runtime loading.

Provides guidance on using Burn deep learning framework for Rust projects. Invoke when working with tensor operations, model inference, or ONNX model integration in Rust.

Burn WGPU backend - Cross-platform GPU acceleration using WebGPU. Supports Vulkan, Metal, Direct3D, and WebGL.

Review one mvsep-rs rewrite batch quality gate without writing production code. Use when the user asks for behavior parity review, error tracing review, async or ergonomics review, data structure or algorithm review, Rust style audit, frontend UX audit, or promotion readiness for a mvsep-rs migration batch.

Expand dependency sources and align canonical shared capabilities, Rust crate reuse, adapters, hand-written Burn gaps, fixtures, and bootstrap records for Vocal2Midi. Use before implementation when imports, native/FFI sources, dependency mismatch, shared model prerequisites, temporary dependency copies, Cargo coordination, or manifest re-cut decisions affect a unit.

Review one Vocal2Midi Rust migration unit quality gate without writing production code. Use for Python/Rust behavior parity, dependency/seam, error tracing, data/algorithm, Rust style, architecture, ergonomics, or promotion readiness.

Coordinate the Vocal2Midi Python-to-Rust rewrite. Use when the user asks to continue, plan, route, re-cut, shard, implement, audit, promote, choose serial or coordinated-parallel execution, or select the next migration unit; when shared Burn gaps, dependency source expansion, capability coverage, or workflow may change the manifest inventory.

Implement exactly one confirmed Vocal2Midi Rust migration unit as the writer role. Use when dependency discovery has accepted the boundary, canonical dependencies, and legacy public behavior seam, and the user asks to add fixtures, port behavior, or prepare one unit for review. Do not use for review-only or shared dependency coordination requests.

[DRAFT] 消费 crate reconnaissance 或用户手动生态证据，按 NOTES.md 中的偏好对齐 Python/Rust 或 legacy/Rust 依赖。覆盖 canonical shared dependencies、crate/backend/adapter/手写能力组合、manifest 重切、seam、验证和桥接；用于避免串行或并行 writer 创建冲突依赖副本。

[DRAFT] 跑通迁移 seam 的最小环境。根据项目选择 FFI、CLI、service、Tauri command、library 或 pipeline adapter，并证明参数、返回值、错误、日志和并发边界可用。

[DRAFT] 第 0 轮行为一致性门禁。对一个迁移单元或已收批的 review batch，独立证明旧实现与新实现在选定 public seam 后行为一致；不写生产代码，只运行/补充对比证据并输出报告。

Prepare dependency, asset, tensor-interface, seam, and rollback evidence for one named Vocal2Midi Linux model unit under the py2rs control plane.

Review exactly one unit and one declared role in one named Vocal2Midi Linux model manifest without writing production code.

Coordinate one named Vocal2Midi Linux model shard under the py2rs model control plane. Use when selecting, re-cutting, routing, implementing, reviewing, or promoting Qwen3-ASR, Romaji ASR, HubertFA, GAME, or RMVPE work.

Implement exactly one confirmed unit from one named Vocal2Midi model manifest as the writer role after dependency/bootstrap acceptance.

Implement one mvsep-rs backend rewrite batch as the code-writing role. Use when the user asks to write code for a selected migration batch, create AppBackend seams, move config/formats/algorithm cache/transfer/tasks to the rewritten backend, or prepare a batch for independent review. Do not use for review-only requests.

Review one mvsep-rs rewrite batch quality gate without writing production code. Use when the user asks for behavior parity review, error tracing review, async or ergonomics review, data structure or algorithm review, Rust style audit, frontend UX audit, or promotion readiness for a mvsep-rs migration batch.

Coordinate the mvsep-rs gradual backend rewrite. Use when the user asks to continue, plan, implement, audit, or route work for replacing the old Tauri backend with the rewritten Rust backend; when updating migration state, project constraints, rewrite docs, or multi-agent rewrite workflow; or when deciding the next mvsep-rs migration batch.