convex-performance-audit

The `convex-performance-audit` skill is designed to diagnose and optimize performance issues in Convex applications. It systematically analyzes problem areas such as hot-path reads, write contention, subscription costs, and function execution limits. By using available deployment health signals, CLI insights, or runtime observations, this skill helps developers identify and address inefficiencies that can cause slow responses, high resource usage, or repeated errors in Convex apps. It focuses on actionable guidance rather than unnecessary refactoring, ensuring that optimizations are targeted and evidence-based.

Key features of this skill include identifying high byte or document reads, OCC conflict errors, excessive mutation retries, subscription inefficiencies, slow UI updates, and functions nearing execution or transaction limits. It provides structured references for each problem class, including best practices for analyzing and mitigating these issues. The skill enforces guardrails to avoid overly complex solutions unless there is a measurable signal, and it recommends escalation for invasive or migration-heavy fixes. It supports gathering signals from Convex CLI insights, existing doctor tools, or code audits, enabling a flexible approach to performance troubleshooting.

This skill is ideal for developers, engineers, and performance specialists working with Convex applications who need to improve application responsiveness, reduce resource consumption, and prevent latency or read amplification problems. It is particularly useful when users encounter OCC conflicts, high subscription costs, or function execution limits, or when deployment insights indicate performance bottlenecks. It is not intended for initial setup, pure schema migrations, or minor one-off optimizations without clear performance impact. By guiding users through problem identification, signal analysis, and structured reference materials, this skill helps maintain high-performing Convex apps at scale.