Developing Genkit Go

This wraps Azure API Management around your AI models to add governance policies like semantic caching, token limits, and content safety filters. You'd use it when you need to control costs (caching can cut bills 60-80%), prevent jailbreaks, or rate-limit your MCP tools before they hit production. It generates the APIM policies, sets up managed identity auth to Azure OpenAI, and gives you the curl commands to test everything. The policy examples are solid and it covers the tricky bits like proper policy ordering. Honestly feels like something Microsoft should have built into the Azure portal by now, but this gets you there with CLI commands.

Forces Claude to actually think before coding, which frankly should be the default but isn't. Blocks all implementation until you've walked through context exploration, clarifying questions, approach options, and design approval. Generates proper spec docs with self-review checks for placeholders and contradictions. The hard gate against "this is too simple to design" saves you from those deceptively basic requests that turn into scope creep nightmares. Most useful when you're tired of Claude jumping straight into half-baked solutions that miss obvious requirements or architectural decisions.

Azure-aigateway enables developers and platform engineers to configure Azure API Management as a governance layer for AI models, MCP tools, and agents. The skill provides policies and commands for implementing semantic caching, token limits, content safety filters, load balancing, rate limiting, and jailbreak detection to control costs, ensure compliance, and protect AI workloads. It solves the problem of managing multiple AI backends and tools through a single controlled gateway with observability and cost control mechanisms.

A solid reference for anyone working with Postgres, especially on Supabase. Contains 8 categories of performance rules prioritized by impact, from critical query optimization to advanced features. Each rule includes wrong vs. right SQL examples, query plan analysis, and specific performance metrics. The prioritization system is smart - it focuses on the stuff that actually breaks production first (missing indexes, connection pooling) before getting into nice-to-haves. Worth having loaded when reviewing database code or designing schemas, since it catches common Postgres pitfalls that can tank performance.

When Lark's existing CLI commands don't cover what you need, this skill digs into the official documentation to find raw OpenAPI endpoints you can call directly. It walks through the three-layer doc structure (index → module → specific API), extracts the complete specs including parameters and permissions, then shows you the exact lark-cli api command to run. Honestly, it's a bit tedious having to crawl through markdown files, but it beats manually hunting through Lark's documentation when you need that one specific API that isn't wrapped yet. Good for power users who hit CLI limitations.

When your UI starts drifting from your design system, this skill audits everything and brings it back into alignment. It systematically replaces hard-coded values with design tokens, swaps custom components for design system equivalents, and fixes spacing, typography, and color inconsistencies. The preparation phase is thorough, requiring you to invoke /frontend-design first and study your design system documentation before making changes. It's particularly effective when you inherit messy code or notice visual inconsistencies creeping into features. The cleanup is comprehensive, removing orphaned styles and consolidating duplicate components.