convex-suggest

Suggests matching Convex components when code duplicates existing component functionality.

9.4k|1.5k|Updated Jan 3, 2026
One-click install
npx skills add https://github.com/openclaw/clawhub --skill convex-suggest
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: convex-suggest
Source: https://github.com/openclaw/clawhub/tree/main/.agents/skills/convex-suggest
Command: npx skills add https://github.com/openclaw/clawhub --skill convex-suggest

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Developers often hand-roll patterns like crons, rate limiters, counters, or search that Convex components already solve, wasting effort and introducing scaling bugs. This Skill passively detects those patterns and suggests the right component after the task is done.

Core Features & Use Cases

  • Pattern Detection: Matches code and intent against detector rules covering crons, sharded-counter, rate-limiter, storage, search, presence, workflow, RAG, and prosemirror-sync.
  • Passive, Non-Interrupting Suggestions: Surfaces exactly one suggestion after completing the current task, citing the specific code that triggered it.
  • Consent-Gated Installation: Never installs a component without explicit user approval, and never repeats a dismissed suggestion.
  • Use Case: While writing a mutation that increments post.likes + 1 under concurrent access, the Skill finishes your task, then suggests @convex-dev/sharded-counter and explains the OCC conflict risk in your own code.

Quick Start

Ask the assistant to review your Convex backend code and suggest any existing components that could replace hand-rolled patterns.

Frequently Asked Questions about convex-suggest

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I know which Convex component to use for my backend pattern?

The Skill matches your code against detector rules: setInterval or cron logic maps to @convex-dev/crons, shared counter increments to @convex-dev/sharded-counter, rate-limit counters to @convex-dev/rate-limiter, and external vector databases to @convex-dev/rag. It names one component and explains why it fits your code.

How does the Convex component suggestion workflow work?

It passively observes your code and requests, finishes the current task first, then offers one suggestion quoting the triggering code. If you agree, it runs the install hint such as /add <component>; if you decline, it drops the suggestion permanently.

Will it install Convex components automatically without asking?

No. The Skill never installs anything without explicit consent. It suggests the component, explains the reasoning anchored in your code, and waits for a yes before running any install command.

What patterns does Convex component detection cover?

Detection covers email/SMTP, push notifications, scheduled jobs, shared counters, aggregation scans, multi-step actions, bounded concurrency, rate limiting, file uploads, full-text search, presence indicators, vector search, and collaborative editing with prosemirror-sync.

When does the suggestion Skill not trigger?

It does not fire on generic coding questions unrelated to Convex, such as sorting arrays or writing CSS. It also skips components you have already installed and never repeats a suggestion you declined or ignored.