Browse reusable Agent Skills, each with a clear purpose and practical guidance.
How to add/upload tuned config CSVs under aiter/configs (incl. model_configs/) without introducing duplicate shapes, and how to find & resolve duplicate-shape collisions. Use whenever adding a model's tuned config, merging/uploading config CSVs, editing anything under aiter/configs/**, or when a run hits "duplicate shape entries during merge".
Create a complete, accurate DSC resource in this repository following the provided guidelines and design patterns.
Generate comprehensive technical design translating requirements (WHAT) into architecture (HOW) with discovery process. Use when creating architecture from requirements.
Build onchain apps with Coinbase's OnchainKit React components - wallets, swaps, NFTs, payments
Compile-time optimization guidance for HIP/C++ kernels using opus.hpp. Use when writing or reviewing OPUS kernels, analyzing compile time, reducing template instantiation overhead, or optimizing hipcc build performance.
Module-level JIT build-wall optimization for opus-based aiter modules. Use when an aiter JIT module's first-call build wall is a user-visible bottleneck or when adding a new module.
Build and harden React Flow diagram surfaces in VeloxDB using @xyflow/react patterns (nodes, edges, viewport, controls, interactions, performance). Use when working on schema diagram canvas behavior, replacing legacy canvas code, or improving diagram UX and scale.
Enforces a Rust-first approach for heavy work in VeloxDB so expensive compute stays in Tauri/Rust instead of JavaScript. Focuses on memory efficiency, fast execution, bounded IPC payloads, and simple implementation patterns. Use when adding or changing heavy queries, transformations, diffing, parsing, exports, aggregations, or any path where performance is critical.
Guides scalability and performance work for VeloxDB's Tauri + Rust PostgreSQL backend and React + TanStack frontend. Covers pnpm-only tooling, src/ file layout (features, data, components/ui, lib), connection pooling, bounded query results, IPC payload size, TanStack Query defaults, virtualization, memo patterns, minimal useEffect usage, and shadcn-style UI consistency. Use when the user asks about performance, memory, latency, large result sets, slow queries, pooling, caching, N+1 patterns, scaling the UI, effects/hooks, repo structure, package manager, or component styling aligned with the design system.