CO

convex-functions

It formats Convex backend operations while enforcing linting rules and argument validation requirements.

Install

mkdir -p .claude/skills/convex-functions && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/2586" && unzip -o skill.zip -d .claude/skills/convex-functions && rm skill.zip

Installs to .claude/skills/convex-functions

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Writing queries, mutations, actions, and HTTP actions with proper argument validation, error handling, internal functions, and runtime considerations
149 charsno explicit “when” trigger
Intermediate

Key capabilities

  • →Structure queries, mutations, and actions
  • →Implement argument validation using Convex validators
  • →Manage database operations with explicit table names
  • →Schedule functions for delayed execution
  • →Handle HTTP webhooks and API endpoints

How it works

It organizes backend logic into reactive queries, transactional mutations, and external-facing actions, all enforced by schema-based validators.

Inputs & outputs

You give it
Function definition with arguments
You get back
Validated database operation or API response

When to use convex-functions

  • →Create a read-only query function
  • →Implement a database mutation
  • →Set up an HTTP action

About this skill

Convex functions

Every exported function in convex/ uses the object form with args and returns validators. Pick the type by what the handler touches: queries read, mutations write, actions call out.

Pick the function type

TypeDatabaseExternal callsCallable byUse for
queryReadNoClients, other functionsReads. Cached and reactive.
mutationRead and writeNoClients, other functionsWrites. One transaction.
actionOnly via runQuery and runMutationYesClients, scheduler, other actionsfetch, third party SDKs, Node APIs
internalQuery, internalMutation, internalActionSame as the public formSameOnly other Convex functionsScheduled work, crons, privileged writes
httpActionOnly via runQuery and runMutationYesHTTP requests in convex/http.tsWebhooks, REST endpoints

Default to query or mutation. Reach for an action only when the handler must talk to something outside Convex.

The object form

Declare args and returns on every function. A function that returns nothing declares returns: v.null() and returns null. Hoist a shared document validator when several functions return the same shape.

// convex/tasks.ts
import { query, mutation } from "./_generated/server";
import { v } from "convex/values";

const taskValidator = v.object({
  _id: v.id("tasks"),
  _creationTime: v.number(),
  userId: v.id("users"),
  title: v.string(),
  completed: v.boolean(),
});

export const get = query({
  args: { taskId: v.id("tasks") },
  returns: v.union(taskValidator, v.null()),
  handler: async (ctx, args) => {
    return await ctx.db.get(args.taskId);
  },
});

export const remove = mutation({
  args: { taskId: v.id("tasks") },
  returns: v.null(),
  handler: async (ctx, args) => {
    await ctx.db.delete(args.taskId);
    return null;
  },
});

Reading data

Use ctx.db.get(id) for one document by id. For everything else use withIndex against an index defined in convex/schema.ts. Never call .filter() on a table query; it scans the whole table.

export const listByUser = query({
  args: { userId: v.id("users") },
  returns: v.array(taskValidator),
  handler: async (ctx, args) => {
    return await ctx.db
      .query("tasks")
      .withIndex("by_user", (q) => q.eq("userId", args.userId))
      .order("desc")
      .take(50);
  },
});

Pick the terminal method by how many documents you expect:

MethodReturnsUse when
.unique()One doc or null, throws on more than oneThe index guarantees at most one match
.first()First doc or nullYou want the newest or oldest match
.take(n)Up to n docsA bounded list such as a recent feed
.collect()Every matchThe result set is small and stays small
.paginate(opts)A page plus cursorThe table is unbounded

Paginated queries take paginationOpts: paginationOptsValidator (from convex/server) as an argument.

Writing data

MethodWhat it does
ctx.db.insert("tasks", doc)Inserts and returns the new id
ctx.db.patch(id, fields)Shallow merges fields. Throws if the doc is missing
ctx.db.replace(id, doc)Replaces the whole doc. Throws if missing
ctx.db.delete(id)Deletes the doc

Patch directly when you do not need the old value. Reading first widens the window for write conflicts. Make mutations safe to retry.

export const rename = mutation({
  args: { taskId: v.id("tasks"), title: v.string() },
  returns: v.null(),
  handler: async (ctx, args) => {
    await ctx.db.patch(args.taskId, { title: args.title });
    return null;
  },
});

Internal functions and references

query, mutation, and action are public. Anyone with the deployment URL can call them. Use internalQuery, internalMutation, and internalAction for code that should only run from other Convex code: scheduled jobs, crons, webhook handlers, privileged writes.

Reference functions through the generated objects in ./_generated/api:

  • api.tasks.get points at a public function in convex/tasks.ts
  • internal.tasks.markPaid points at an internal function in the same file
  • Folders map to paths: convex/billing/invoices.ts gives api.billing.invoices.list

Always schedule internal.*. Scheduled functions and crons run without a client, so a public reference there skips the auth checks a client call would hit.

// convex/messages.ts
import { mutation, internalMutation } from "./_generated/server";
import { internal } from "./_generated/api";
import { v } from "convex/values";

export const send = mutation({
  args: { channelId: v.id("channels"), content: v.string() },
  returns: v.id("messages"),
  handler: async (ctx, args) => {
    const messageId = await ctx.db.insert("messages", args);
    await ctx.scheduler.runAfter(0, internal.messages.notifySubscribers, {
      channelId: args.channelId,
      messageId,
    });
    return messageId;
  },
});

export const notifySubscribers = internalMutation({
  args: { channelId: v.id("channels"), messageId: v.id("messages") },
  returns: v.null(),
  handler: async (ctx, args) => {
    const subs = await ctx.db
      .query("subscriptions")
      .withIndex("by_channel", (q) => q.eq("channelId", args.channelId))
      .collect();
    await Promise.all(
      subs.map((sub) =>
        ctx.db.insert("notifications", {
          userId: sub.userId,
          messageId: args.messageId,
          read: false,
        }),
      ),
    );
    return null;
  },
});

Actions and runtime boundaries

Actions have no ctx.db. They read through ctx.runQuery and write through ctx.runMutation. Each call is its own transaction, so keep the count low and do related reads and writes inside one mutation.

fetch works in the default runtime. Add "use node"; as the first line of a file only when an action needs Node built ins or a Node only SDK. A "use node" file can export actions only; queries and mutations go in a separate file.

// convex/orders.ts (default runtime)
import { action } from "./_generated/server";
import { internal } from "./_generated/api";
import { v, ConvexError } from "convex/values";
import { Doc } from "./_generated/dataModel";

export const charge = action({
  args: { orderId: v.id("orders") },
  returns: v.null(),
  handler: async (ctx, args) => {
    // Same file call: annotate the result so TypeScript does not hit a circular type
    const order: Doc<"orders"> | null = await ctx.runQuery(
      internal.orders.getForCharge,
      { orderId: args.orderId },
    );
    if (!order) {
      throw new ConvexError("Order not found");
    }
    const res = await fetch("https://api.payments.example/charge", {
      method: "POST",
      body: JSON.stringify({ amount: order.total }),
    });
    await ctx.runMutation(internal.orders.setStatus, {
      orderId: args.orderId,
      status: res.ok ? "paid" : "failed",
    });
    return null;
  },
});

Doc and Id come from ./_generated/dataModel. The annotation is only needed when the called function lives in the same file.

Errors

Throw ConvexError from convex/values for anything a client should read. Its data reaches the client; a plain Error message is redacted in production. Return null for expected absences such as a lookup that finds nothing. Throw for real failures: not authenticated, not authorized, invalid input.

import { ConvexError } from "convex/values";

throw new ConvexError({ code: "NOT_FOUND", message: "Task not found" });

Thin wrappers

Keep handlers short. Put auth lookups, validation, and business logic in plain async functions that take ctx first, then call them from the wrapper. Plain helpers are testable and shared between queries and mutations without a ctx.runQuery hop.

import { QueryCtx, MutationCtx } from "./_generated/server";
import { ConvexError } from "convex/values";

export async function getCurrentUser(ctx: QueryCtx | MutationCtx) {
  const identity = await ctx.auth.getUserIdentity();
  if (!identity) {
    throw new ConvexError("Not authenticated");
  }
  const user = await ctx.db
    .query("users")
    .withIndex("by_token", (q) =>
      q.eq("tokenIdentifier", identity.tokenIdentifier),
    )
    .unique();
  if (!user) {
    throw new ConvexError("User not found");
  }
  return user;
}

From a query or mutation, call the helper directly. ctx.runQuery and ctx.runMutation are for actions and component boundaries.

Common mistakes

MistakeWhy it breaksDo instead
No returns validatorReturn shape drifts and client types lieDeclare returns, use v.null() for nothing
.filter() on a table queryFull table scanAdd an index, use withIndex
ctx.db inside an actionActions have no database handlectx.runQuery and ctx.runMutation
fetch inside a query or mutationTransactions must be deterministicMove it to an action
Scheduling api.*Runs public code without a client, skips authSchedule internal.*
"use node" in a file with queriesBundler rejects the fileSplit actions into their own file
Date.now() in a queryBreaks caching and reactivityPass time as an arg or store a status field
Many runQuery calls from one actionEach is a separate transaction, races appearOne mutation that does the related work
Plain Error for user messagesMessage is hidden in productionConvexError
Missing await on ctx.db or schedulerWrite may not commitAwait every ctx call

Checklist

  • Object form with args and returns on every exported function
  • returns: v.null() and return null when there is nothing to return
  • Reads use ctx.db.get(id) or withIndex, never .filter()
  • [

Content truncated.

When not to use it

  • →Non-Convex backend development
  • →Client-side only state management

Prerequisites

Convex project environmentESLint plugin for Convex

Limitations

  • →Actions cannot access the database directly
  • →Requires specific file structure for generated server code

How it compares

It enforces strict linting and validation rules specific to the Convex runtime, preventing common errors found in generic backend implementations.

Compared to similar skills

convex-functions side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
convex-functions (this skill)47moNo flagsIntermediate
add-league06moReviewIntermediate
nanoclaw-backend-ts04moNo flagsAdvanced
dev-supabase04moReviewIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry