Agent skill
convex
Expert guidance for Convex backend development including queries, mutations, actions, schemas, authentication, scheduling, file storage, search, and Next.js integration. Use when working with Convex functions, database operations, convex/ directory code, or Next.js App Router with Convex. Triggers: convex functions, ctx.db, useQuery, useMutation, usePreloadedQuery, preloadQuery, fetchQuery, convex schema, convex auth, convex cron, convex actions, convex scheduling, ctx.storage, generateUploadUrl, file upload, storage.store, storage.getUrl, Id<"_storage">, ConvexProvider, ConvexClientProvider, db.system, _scheduled_functions, _storage system table, OCC, optimistic concurrency, transaction atomicity, schema evolution, searchIndex, vectorIndex, withSearchIndex, ctx.vectorSearch, full-text search, vector search, embeddings, RAG, semantic search, typeahead search, ConvexError, httpAction, httpRouter, CORS, webhook.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/convex-skill
SKILL.md
Convex Backend Development
Core Architecture
Convex is a reactive database where queries are TypeScript functions. The sync engine (queries + mutations + database) is the heart of Convex — center your app around it.
Function Types
| Type | DB Access | Deterministic | Cached/Reactive | Use For |
|---|---|---|---|---|
query |
Read only | Yes | Yes | All reads, subscriptions |
mutation |
Read/Write | Yes | No | All writes (transactions) |
action |
Via ctx.run* | No | No | External APIs, LLMs, email |
httpAction |
Via ctx.run* | No | No | Webhooks, custom HTTP |
Key rule: Queries and mutations cannot make network requests. Actions cannot directly access the database.
Project Structure (Best Practice)
convex/
├── _generated/ # Auto-generated types (commit this)
├── schema.ts # Database schema
├── model/ # Helper functions (most logic lives here)
│ ├── users.ts
│ └── messages.ts
├── users.ts # Thin wrappers exposing public API
├── messages.ts
├── crons.ts # Cron job definitions
└── http.ts # HTTP action routes
Essential Patterns
1. Function Structure
// convex/messages.ts
import { query, mutation, internalMutation } from './_generated/server';
import { internal } from './_generated/api';
import { v } from 'convex/values';
// PUBLIC query with validators (always validate public functions)
export const list = query({
args: { channelId: v.id('channels') },
handler: async (ctx, { channelId }) => {
return await ctx.db
.query('messages')
.withIndex('by_channel', (q) => q.eq('channelId', channelId))
.order('desc')
.take(50);
}
});
// PUBLIC mutation with validators and auth check
export const send = mutation({
args: { channelId: v.id('channels'), body: v.string() },
handler: async (ctx, { channelId, body }) => {
const user = await ctx.auth.getUserIdentity();
if (!user) throw new Error('Unauthorized');
await ctx.db.insert('messages', {
channelId,
body,
authorId: user.subject
});
}
});
// INTERNAL mutation (for scheduling, crons, actions)
export const deleteOld = internalMutation({
args: { before: v.number() },
handler: async (ctx, { before }) => {
const old = await ctx.db
.query('messages')
.withIndex('by_createdAt', (q) => q.lt('_creationTime', before))
.take(100);
for (const msg of old) {
await ctx.db.delete(msg._id);
}
}
});
2. Helper Functions Pattern
Most logic should live in helper functions, NOT in query/mutation handlers:
// convex/model/users.ts
import { QueryCtx, MutationCtx } from '../_generated/server';
import { Doc } from '../_generated/dataModel';
export async function getCurrentUser(
ctx: QueryCtx
): Promise<Doc<'users'> | null> {
const identity = await ctx.auth.getUserIdentity();
if (!identity) return null;
return await ctx.db
.query('users')
.withIndex('by_tokenIdentifier', (q) =>
q.eq('tokenIdentifier', identity.tokenIdentifier)
)
.unique();
}
export async function requireUser(ctx: QueryCtx): Promise<Doc<'users'>> {
const user = await getCurrentUser(ctx);
if (!user) throw new Error('Unauthorized');
return user;
}
3. Actions with Scheduling
// convex/ai.ts
import { action, internalMutation } from './_generated/server';
import { internal } from './_generated/api';
import { v } from 'convex/values';
export const summarize = action({
args: { documentId: v.id('documents') },
handler: async (ctx, { documentId }) => {
// Read data via internal query
const doc = await ctx.runQuery(internal.documents.get, { documentId });
// Call external API
const response = await fetch('https://api.openai.com/v1/...', {...});
const summary = await response.json();
// Write result via internal mutation
await ctx.runMutation(internal.documents.setSummary, {
documentId,
summary: summary.text
});
}
});
// Trigger action from mutation (not directly from client)
export const requestSummary = mutation({
args: { documentId: v.id('documents') },
handler: async (ctx, { documentId }) => {
const user = await ctx.auth.getUserIdentity();
if (!user) throw new Error('Unauthorized');
await ctx.db.patch(documentId, { status: 'processing' });
// Schedule action (runs after mutation commits)
await ctx.scheduler.runAfter(0, internal.ai.summarizeInternal, {
documentId
});
}
});
4. Application Errors
import { ConvexError } from 'convex/values';
export const assignRole = mutation({
args: { roleId: v.id('roles'), userId: v.id('users') },
handler: async (ctx, { roleId, userId }) => {
const existing = await ctx.db
.query('assignments')
.withIndex('by_role', (q) => q.eq('roleId', roleId))
.first();
if (existing) {
throw new ConvexError({
code: 'ROLE_TAKEN',
message: 'Role is already assigned'
});
}
await ctx.db.insert('assignments', { roleId, userId });
}
});
Critical Rules
DO ✅
- Use
internal.functions for allctx.run*,ctx.scheduler, and crons - Always validate args for public functions with
v.*validators - Always check auth in public functions:
ctx.auth.getUserIdentity() - Use indexes with
.withIndex()instead of.filter() - Await all promises (enable
no-floating-promisesESLint rule) - Keep actions small — put logic in queries/mutations
- Batch database operations in single mutations
- Use
ConvexErrorfor user-facing errors
DON'T ❌
- Don't use
api.functions for scheduling (useinternal.) - Don't use
.filter()on queries — use indexes or TypeScript filter - Don't use
.collect()on unbounded queries (use.take()or pagination) - Don't use
Date.now()in queries (breaks caching) - Don't call actions directly from client (trigger via mutation + scheduler)
- Don't make sequential
ctx.runQuery/runMutationcalls in actions (batch them) - Don't use
ctx.runActionunless switching runtimes (use helper functions)
Reference Guides
For detailed patterns, see:
- FUNCTIONS.md — Queries, mutations, actions, internal functions
- VALIDATION.md — Argument validation, extended validators
- ERROR_HANDLING.md — ConvexError, application errors
- HTTP_ACTIONS.md — HTTP actions, CORS, webhooks
- RUNTIMES.md — Default vs Node.js runtime, bundling, debugging
- DATABASE.md — Schema, indexes, reading/writing data
- SEARCH.md — Full-text search, vector search, RAG patterns
- ADVANCED.md — System tables, schema philosophy, OCC
- AUTH.md — Row-level security, Convex Auth (first-party)
- SCHEDULING.md — Crons, scheduled functions, workflows
- FILE_STORAGE.md — Upload, store, serve, delete files
- NEXTJS.md — Next.js App Router, SSR, Server Actions
Auth Provider Skills (add to project as needed):
convex-auth— Universal auth patterns, storing users, debuggingconvex-clerk— Clerk setup, webhooks, JWT configurationconvex-workos— WorkOS AuthKit setup, auto-provisioning
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?