Skill UI

A comprehensive guide to implementing Perplexity's API integration at various scales. It provides three validated architectural blueprints—Direct Widget, Cached Research Layer, and Multi-Query Pipeline—allowing developers to choose the optimal structure based on required query volume, latency goals, and complexity. Learn how to scale from simple search widgets to automated, high-volume research pipelines.

Guides adding PyTorch FSDP2’s fully_shard integration to training loops, covering correct distributed init, bottom-up sharding, mixed precision/offload policies, DTensor-aware optimizers, and distributed checkpointing so oversized models or DeviceMesh sharding goals work reliably.

This skill provides detailed guidance on authoring effective `specs.md` files for autonomous development workflows. It teaches users how to clearly define sprint goals, establish explicit scope boundaries (In/Out of Scope), and configure rigorous testing requirements (QA, UI Testing). This ensures that agentic development remains focused and prevents scope creep.

Autoresearch Agent runs an autonomous optimization loop that edits a target file, invokes a provided evaluation command, keeps metric-improving commits, drops failures, and repeats indefinitely so engineers can optimize speed, bundle size, prompts, or other measurable goals overnight.

Provides an expert Spring Boot 4 testing guide that explains the test pyramid, slice annotations, AssertJ-centric assertions, MockMvcTester, RestTestClient, Testcontainers, and practical coverage goals.

This tool enforces four rigorous coding principles, inspired by Andrej Karpathy's insights on LLM pitfalls. It guides developers to surface assumptions, prioritize simplicity, make surgical changes, and define verifiable goals. Use it when reviewing code, checking diffs, or before committing to ensure high code quality and prevent over-engineering.

A tool designed for persistent context management within AI coding sessions. Memory Kit solves the problem of context loss that occurs during compaction or session restarts by saving, loading, updating, and sharing critical session state to a dedicated `MEMORY.md` file. Key functions include creating snapshots of goals and decisions, syncing context across team members via Git, and auditing/pruning stale entries, ensuring continuous workflow and knowledge retention.

Alex converts high-level requirements into a precise, dependency-aware, and actionable implementation checklist. It maps critical paths, breaks down features into atomic micro-tasks, and establishes clear Definitions of Done (DoD) across various layers. Ideal for bridging the gap between conceptual goals and step-by-step engineering execution.

This skill applies a rigorous, evidence-first discipline to coding tasks, inspired by Fable-like best practices. It guides agents through a structured workflow: inspecting the workspace, tracking goals and findings, grounding all conclusions in verifiable evidence, and executing thorough verification gates before declaring completion. Ideal for complex development, debugging, or adapting cross-model prompt guidance into robust, verifiable project rules.

This skill guides users through an in-depth interview process to achieve maximum project context completeness (100% "Gold Code") for their .faf file. By systematically asking questions about project goals, architecture, and technology stack, it ensures all critical context slots are filled, making the project fully structured and highly readable by AI systems.

This meta-skill acts as a routing layer for complex UI development tasks. It guides the process of selecting the smallest, most specific, and useful skill context using the ui-skills CLI. Use it when starting UI work to analyze goals, identify the correct category, and pinpoint the precise skill set needed for implementation, ensuring development efficiency and accuracy.

The Agent Harness provides a robust, verifiable loop structure for driving complex goals across specialized domains. It converts a high-level objective into a verifiable task plan, executes tasks sequentially using domain-specific tools, and rigorously verifies every step with machine-run checks. The system includes mechanisms for retries, budget management, and mandatory human escalation, ensuring no goal is closed until all progress is verifiably complete or explicitly waived.