convex-create-component — Claude Code skill: reusable Convex
A skill to design and build reusable Convex components with isolated tables and a clear API boundary. Useful when you need to extract backend logic into a self-contained module or package an integration that owns its own data.
npx skills add convex/agent-skills --skill convex-create-component -a claude-codeWhat convex-create-component does
The skill walks you through the entire process: choosing a component shape, reading the right reference docs (for local, packaged, or hybrid components), and deciding whether a component is actually justified. It guides you with questions and validation at each step.
Inside, there's a structured workflow: define tables and public functions, set up data boundaries, configure convex.config.ts, and generate file structure. The skill knows that authentication and environment access stay in the app, that parent app IDs are passed as strings, and how to correctly import from the component's ./_generated/server.
Once the component is scaffolded, the skill helps wire it into the app with app.use(...), create wrapper functions for clients, and run npx convex dev to validate types and codegen. Works with Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, and OpenCode.
What it helps with
- extract backend logic into a reusable module
- build a Convex component with its own tables and API
- structure an integration that owns its persistent state
- set clear boundaries between component and app code
- package functionality for reuse across multiple apps
Example prompts for the agent
- Design a Convex component for notifications with a notifications table and send and listUnread functions
- Help me split authentication logic into a separate component and pick the right architecture shape
- Create a local component for payments handling and set up boundaries and wrapper functions correctly
How to install convex-create-component
1Skills CLI (Vercel)
npx skills add convex/agent-skills --skill convex-create-component -a claude-code2Manually: put the files in place
~/.claude/skills/convex-create-component/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/convex-create-component/
1Skills CLI (Vercel)
npx skills add convex/agent-skills --skill convex-create-component -a codex2Manually: put the files in place
.agents/skills/convex-create-component/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add convex/agent-skills --skill convex-create-component -a cursor2Manually: put the files in place
.cursor/skills/convex-create-component/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install convex/agent-skills convex-create-component2Manually: put the files in place
.github/skills/convex-create-component/SKILL.md1Skills CLI (Vercel)
npx skills add convex/agent-skills --skill convex-create-component -a gemini-cli2Platform CLI
gemini skills install https://github.com/convex/agent-skills --path skills/convex-create-componentManually: put the files in place
.devin/skills/convex-create-component/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add convex/agent-skills --skill convex-create-component -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Similar
Questions about convex-create-component
How do I install convex-create-component in Claude Code?
Run npx skills add convex/agent-skills --skill convex-create-component -a claude-code. Then check ~/.claude/skills/convex-create-component/SKILL.md.
What does convex-create-component do?
The skill helps you design and build a reusable Convex component: define tables and API surface, establish boundaries between component and app, wire it up with app.use(...), and validate types.
When should I use a Convex component vs. regular app code?
A component is worth it when you need isolated tables, backend functions, or reusable persistent state. For one-off logic, thin utilities, or app-level orchestration, stick with normal code or a TypeScript library.
Source text
SKILL.md · Apache-2.0
---
name: convex-create-component
description:
Builds reusable Convex components with isolated tables and app-facing APIs.
Use for new components, reusable backend modules, integrations, or component
boundary work.
---
# Convex Create Component
Create reusable Convex components with clear boundaries and a small app-facing
API.
## When to Use
- Creating a new Convex component in an existing app
- Extracting reusable backend logic into a component
- Building a third-party integration that should own its own tables and
workflows
- Packaging Convex functionality for reuse across multiple apps
## When Not to Use
- One-off business logic that belongs in the main app
- Thin utilities that do not need Convex tables or functions
- App-level orchestration that should stay in `convex/`
- Cases where a normal TypeScript library is enough
## Workflow
1. Ask the user what they are building and what the end goal is. If the repo
already makes the answer obvious, say so and confirm before proceeding.
2. Choose the shape using the decision tree below and read the matching
reference file.
3. Decide whether a component is justified. Prefer normal app code or a regular
library if the feature does not need isolated tables, backend functions, or
reusable persistent state.
4. Make a short plan for:
- what tables the component owns
- what public functions it exposes
- what data must be passed in from the app (auth, env vars, parent IDs)
- what stays in the app as wrappers or HTTP mounts
5. Create the component structure with `convex.config.ts`, `schema.ts`, and
function files.
6. Implement functions using the component's own `./_generated/server` imports,
not the app's generated files.
7. Wire the component into the app with `app.use(...)`. If the app does not
already have `convex/convex.config.ts`, create it.
8. Call the component from the app through `components.<name>` using
`ctx.runQuery`, `ctx.runMutation`, or `ctx.runAction`.
9. If React clients, HTTP callers, or public APIs need access, create wrapper
functions in the app instead of exposing component functions directly.
10. Run `npx convex dev` and fix codegen, type, or boundary issues before
finishing.
## Choose the Shape
Ask the user, then pick one path:
| Goal | Shape | Reference |
| ------------------------------------------------- | ---------------- | ----------------------------------- |
| Component for this app only | Local | `references/local-components.md` |
| Publish or share across apps | Packaged | `references/packaged-components.md` |
| User explicitly needs local + shared library code | Hybrid | `references/hybrid-components.md` |
| Not sure | Default to local | `references/local-components.md` |
Read exactly one reference file before proceeding.
## Default Approach
Unless the user explicitly wants an npm package, default to a local component:
- Put it under `convex/components/<componentName>/`
- Define it with `defineComponent(...)` in its own `convex.config.ts`
- Install it from the app's `convex/convex.config.ts` with `app.use(...)`
- Let `npx convex dev` generate the component's own `_generated/` files
## Component Skeleton
A minimal local component with a table and two functions, plus the app wiring.
```ts
// convex/components/notifications/convex.config.ts
import { defineComponent } from "convex/server";
export default defineComponent("notifications");
```
```ts
// convex/components/notifications/schema.ts
import { defineSchema, defineTable } from "convex/server";
import { v } from "convex/values";
export default defineSchema({
notifications: defineTable({
userId: v.string(),
message: v.string(),
read: v.boolean(),
}).index("by_user_read", ["userId", "read"]),
});
```
```ts
// convex/components/notifications/lib.ts
import { v } from "convex/values";
import { mutation, query } from "./_generated/server.js";
export const send = mutation({
args: { userId: v.string(), message: v.string() },
returns: v.id("notifications"),
handler: async (ctx, args) => {
return await ctx.db.insert("notifications", {
userId: args.userId,
message: args.message,
read: false,
});
},
});
export const listUnread = query({
args: { userId: v.string() },
returns: v.array(
v.object({
_id: v.id("notifications"),
_creationTime: v.number(),
userId: v.string(),
message: v.string(),
read: v.boolean(),
}),
),
handler: async (ctx, args) => {
return await ctx.db
.query("notifications")
.withIndex("by_user_read", (q) =>
q.eq("userId", args.userId).eq("read", false),
)
.collect();
},
});
```
```ts
// convex/convex.config.ts
import { defineApp } from "convex/server";
import notifications from "./components/notifications/convex.config.js";
const app = defineApp();
app.use(notifications);
export default app;
```
```ts
// convex/notifications.ts (app-side wrapper)
import { v } from "convex/values";
import { mutation, query } from "./_generated/server";
import { components } from "./_generated/api";
import { getAuthUserId } from "@convex-dev/auth/server";
export const sendNotification = mutation({
args: { message: v.string() },
returns: v.null(),
handler: async (ctx, args) => {
const userId = await getAuthUserId(ctx);
if (!userId) throw new Error("Not authenticated");
await ctx.runMutation(components.notifications.lib.send, {
userId,
message: args.message,
});
return null;
},
});
export const myUnread = query({
args: {},
handler: async (ctx) => {
const userId = await getAuthUserId(ctx);
if (!userId) throw new Error("Not authenticated");
return await ctx.runQuery(components.notifications.lib.listUnread, {
userId,
});
},
});
```
Note the reference path shape: a function in
`convex/components/notifications/lib.ts` is called as
`components.notifications.lib.send` from the app.
## Critical Rules
- Keep authentication in the app, because `ctx.auth` is not available inside
components.
- Keep environment access in the app, because component functions cannot read
`process.env`.
- Pass parent app IDs across the boundary as strings, because `Id` types become
plain strings in the app-facing `ComponentApi`.
- Do not use `v.id("parentTable")` for app-owned tables inside component args or
schema, because the component has no access to the app's table namespace.
- Import `query`, `mutation`, and `action` from the component's own
`./_generated/server`, not the app's generated files.
- Do not expose component functions directly to clients. Create app wrappers
when client access is needed, because components are internal and need
auth/env wiring the app provides.
- If the component defines HTTP handlers, mount the routes in the app's
`convex/http.ts`, because components cannot register their own HTTP routes.
- If the component needs pagination, use `paginator` from `convex-helpers`
instead of built-in `.paginate()`, because `.paginate()` does not work across
the component boundary.
- Define indexes for queried fields instead of using Convex `.filter()` after a
database query.
- Add `args` and `returns` validators to all public component functions, because
the component boundary requires explicit type contracts.
## Patterns
### Authentication and environment access
```ts
// Bad: component code cannot rely on app auth or env
const identity = await ctx.auth.getUserIdentity();
const apiKey = process.env.OPENAI_API_KEY;
```
```ts
// Good: the app resolves auth and env, then passes explicit values
const userId = await getAuthUserId(ctx);
if (!userId) throw new Error("Not authenticated");
await ctx.runAction(components.translator.translate, {
userId,
apiKey: process.env.OPENAI_API_KEY,
text: args.text,
});
```
### Client-facing API
```ts
// Bad: assuming a component function is directly callable by clients
export const send = components.notifications.send;
```
```ts
// Good: re-export through an app mutation or query
export const sendNotification = mutation({
args: { message: v.string() },
returns: v.null(),
handler: async (ctx, args) => {
const userId = await getAuthUserId(ctx);
if (!userId) throw new Error("Not authenticated");
await ctx.runMutation(components.notifications.lib.send, {
userId,
message: args.message,
});
return null;
},
});
```
### IDs across the boundary
```ts
// Bad: parent app table IDs are not valid component validators
args: {
userId: v.id("users"),
}
```
```ts
// Good: treat parent-owned IDs as strings at the boundary
args: {
userId: v.string(),
}
```
### Advanced Patterns
For additional patterns including function handles for callbacks, deriving
validators from schema, static configuration with a globals table, and
class-based client wrappers, see `references/advanced-patterns.md`.
## Validation
Try validation in this order:
1. `npx convex codegen --component-dir convex/components/<name>`
2. `npx convex codegen`
3. `npx convex dev`
Important:
- Fresh repos may fail these commands until `CONVEX_DEPLOYMENT` is configured.
- Until codegen runs, component-local `./_generated/*` imports and app-side
`components.<name>...` references will not typecheck.
- If validation blocks on Convex login or deployment setup, stop and ask the
user for that exact step instead of guessing.
## Reference Files
Read exactly one of these after the user confirms the goal:
- `references/local-components.md`
- `references/packaged-components.md`
- `references/hybrid-components.md`
Official docs:
[Authoring Components](https://docs.convex.dev/components/authoring)
## Checklist
- [ ] Asked the user what they want to build and confirmed the shape
- [ ] Read the matching reference file
- [ ] Confirmed a component is the right abstraction
- [ ] Planned tables, public API, boundaries, and app wrappers
- [ ] Component lives under `convex/components/<name>/` (or package layout if
publishing)
- [ ] Component imports from its own `./_generated/server`
- [ ] Auth, env access, and HTTP routes stay in the app
- [ ] Parent app IDs cross the boundary as `v.string()`
- [ ] Public functions have `args` and `returns` validators
- [ ] Ran `npx convex dev` and fixed codegen or type issues
© convex · Apache-2.0 · convex/agent-skills