skills/instantdb/SKILL.md
Build complete, functional apps with InstantDB as the backend. Use when creating React/vanilla JS or expo applications. Triggers on requests for building apps.
npx skillsauth add instantdb/skills instantdbInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Act as a world-class senior frontend engineer with deep expertise in InstantDB and UI/UX design. Your primary goal is to generate complete and functional apps with excellent visual aesthetics using InstantDB as the backend.
Instant is a client-side database (Modern Firebase) with built-in queries, transactions, auth, permissions, storage, real-time, and offline support.
Instant provides client-side JS SDKs and an admin SDK:
@instantdb/core --- vanilla JS@instantdb/react --- React@instantdb/react-native --- React Native / Expo@instantdb/solidjs --- SolidJS@instantdb/svelte --- Svelte@instantdb/vue --- Vue@instantdb/admin --- backend scripts / serversWhen installing, always check what package manager the project uses (npm, pnpm, bun) first and then install the latest version of the Instant SDK. If working in React use Next and Tailwind unless specified otherwise.
Look for instant.schema.ts and instant.perms.ts. These define the schema and permissions.
Look for an app id and admin token in .env or another env file.
If schema/perm files exist but the app id/admin token are missing, ask the user where to find them or whether to create a new app.
To create a new app:
npx instant-cli init-without-files --title <APP_NAME>
This outputs an app id and admin token. Store them in an env file.
If you get an error related to not being logged in tell the user to:
npx instant-cli login to authenticate the CLIIf you have an app id/admin token but no schema/perm files, pull them:
npx instant-cli pull --yes
Edit instant.schema.ts, then push:
npx instant-cli push schema --yes
New fields = additions; missing fields = deletions.
To rename fields:
npx instant-cli push schema --rename 'posts.author:posts.creator stores.owner:stores.manager' --yes
Edit instant.perms.ts, then push:
npx instant-cli push perms --yes
CRITICAL: When using React make sure to follow the rules of hooks. Remember, you can't have hooks show up conditionally.
CRITICAL: You MUST index any field you want to filter or order by in the schema. If you do not, you will get an error when you try to filter or order by it.
Here is how ordering works:
Ordering: order: { field: 'asc' | 'desc' }
Example: $: { order: { dueDate: 'asc' } }
Notes: - Field must be indexed + typed in schema
- Cannot order by nested attributes (e.g. 'owner.name')
CRITICAL: Here is a concise summary of the where operator map which defines all the filtering options you can use with InstantDB queries to narrow results based on field values, comparisons, arrays, text patterns, and logical conditions.
Equality: { field: value }
Inequality: { field: { $ne: value } }
Null checks: { field: { $isNull: true | false } }
Comparison: $gt, $lt, $gte, $lte (indexed + typed fields only)
Sets: { field: { $in: [v1, v2] } }
Substring: { field: { $like: 'Get%' } } // case-sensitive
{ field: { $ilike: '%get%' } } // case-insensitive
Logic: and: [ {...}, {...} ]
or: [ {...}, {...} ]
Nested fields: 'relation.field': value
CRITICAL: The operator map above is the full set of where filters Instant
supports right now. There is no $exists, $nin, or $regex. And $like and
$ilike are what you use for startsWith / endsWith / includes.
CRITICAL: Pagination keys (limit, offset, first, after, last, before) only work on top-level namespaces. DO NOT use them on nested relations or else you will get an error.
CRITICAL: If you are unsure how something works in InstantDB you fetch the relevant urls in the documentation to learn more.
Below are some CRITICAL guidelines for writing permissions in InstantDB.
data.refdata.ref("<path.to.attr>") for linked attributes.Correct
auth.id in data.ref('post.author.id') // auth.id in list of author ids
data.ref('owner.id') == [] // there is no owner
Errors
auth.id in data.post.author.id
auth.id in data.ref('author')
data.ref('admins.id') == auth.id
auth.id == data.ref('owner.id')
data.ref('owner.id') == null
data.ref('owner.id').length > 0
auth.refdata.ref but path must start with $user.Correct
'admin' in auth.ref('$user.role.type')
auth.ref('$user.role.type')[0] == 'admin'
Errors
auth.ref('role.type')
auth.ref('$user.role.type') == 'admin'
newData.ref('x')
data.ref(someVar + '.members.id')
view permission is auth.id == data.idupdate and delete permissions is falsecreate permission is true (anyone can sign up)view, update, and createdeletecreate rule runs during auth signup flows (not via transact). Use it to restrict signups or validate extraFields.extraFields require an explicit create rule. Without one, signup is blocked to prevent unvalidated writes.data.ref does not work for $files permissions.data.path.startsWith(...) or data.path.endsWith(...) to write
path-based rules.Restrict access to specific fields while keeping the entity public:
{
"$users": {
"allow": {
"view": "true"
},
"fields": {
"email": "auth.id == data.id"
}
}
}
Notes:
view for that fieldCRITICAL: If an app displays images or files, use Instant Storage. Do not store URLs as string attributes on your entities. This includes seed scripts: do not use placeholder image URLs (e.g. picsum.photos) as string attributes to fake file support.
Uploads auto-create $files entities. Link them to your data via the schema,
then query through the relationship to get URLs.
CRITICAL: You MUST include $files in your schema entities if you use Storage.
CRITICAL: $files entities can only be created via db.storage.uploadFile. You
cannot create $files via db.transact, and you cannot set url via transactions.
entities: {
$files: i.entity({
path: i.string().unique().indexed(),
url: i.string(),
}),
posts: i.entity({
caption: i.string(),
}),
},
links: {
postImage: {
forward: { on: "posts", has: "one", label: "image" },
reverse: { on: "$files", has: "many", label: "posts" },
},
}
// Upload and link the returned file ID to your entity
const postId = id();
const { data } = await db.storage.uploadFile(`posts/${postId}/${file.name}`, file);
db.transact(
db.tx.posts[postId].update({ caption }).link({ image: data.id })
);
// Query through the relationship to get the URL
const { data } = db.useQuery({ posts: { image: {} } });
<img src={post.image.url} />
CRITICAL: Hooks for presence and topics live on db.rooms and take the room as the first arg. The room object itself has no usePresence or publishPresence methods.
Rooms host two ephemeral primitives: presence (cursor positions, who's online) and topics (live reactions). Use them only for data that should NOT persist. Persisted data via transact already syncs in real-time to all subscribed clients, so reach for rooms only when the data is intentionally ephemeral.
Each peer publishes a presence object readable by all other peers in the room. Retained for the connection and cleaned up automatically on disconnect.
const room = db.room('chat', 'main');
const { user, peers, publishPresence } = db.rooms.usePresence(room, {
initialPresence: { x: 0, y: 0 },
});
// peers is keyed by peerId, not an array. Use Object.values(peers) to iterate
publishPresence({ x: 50, y: 50 });
Topic payloads aren't retained. Peers only see events fired while they're listening.
const room = db.room('chat', 'main');
const publishEmoji = db.rooms.usePublishTopic(room, 'emoji');
publishEmoji({ name: 'fire' });
db.rooms.useTopicEffect(room, 'emoji', (payload) => {
animateEmoji(payload.name);
});
schema when initializing InstantAlways pass schema when initializing Instant to get type safety for queries and transactions
import schema from '@/instant.schema';
// On client
import { init } from '@instantdb/react'; // or your relevant Instant SDK
const clientDb = init({ appId, schema });
// On backend
import { init } from '@instantdb/admin';
const adminDb = init({ appId, adminToken, schema });
id() to generate idsAlways use id() to generate ids for new entities
import { id } from '@instantdb/react'; // or your relevant Instant SDK
import { clientDb } from '@/lib/clientDb';
clientDb.transact(clientDb.tx.todos[id()].create({ title: 'New Todo' }));
Always use Instant utility types to type data models
import { AppSchema } from '@/instant.schema';
type Todo = InstaQLEntity<AppSchema, 'todos'>; // todo from clientDb.useQuery({ todos: {} })
type PostsWithProfile = InstaQLEntity<
AppSchema,
'posts',
{ author: { avatar: {} } }
>; // post from clientDb.useQuery({ posts: { author: { avatar: {} } } })
db.useAuth or db.subscribeAuth for auth stateimport { clientDb } from '@/lib/clientDb';
// For react/react-native apps use db.useAuth
function App() {
const { isLoading, user, error } = clientDb.useAuth();
if (isLoading) {
return null;
}
if (error) {
return <Error message={error.message} />;
}
if (user) {
return <Main />;
}
return <Login />;
}
// For vanilla JS apps use db.subscribeAuth
function App() {
renderLoading();
db.subscribeAuth((auth) => {
if (auth.error) {
renderAuthError(auth.error.message);
} else if (auth.user) {
renderLoggedInPage(auth.user);
} else {
renderSignInPage();
}
});
}
extraFieldsPass extraFields to any sign-in method to write custom $users properties atomically on user creation.
Fields must be defined as optional attrs on $users in your schema.
Use the created boolean to scaffold data for new users.
// Set properties at signup
const { user, created } = await db.auth.signInWithMagicCode({
email,
code,
extraFields: { nickname, createdAt: Date.now() },
});
// Scaffold data for new users
if (created) {
db.transact([
db.tx.settings[id()]
.update({ theme: 'light', notifications: true })
.link({ user: user.id }),
]);
}
Run npx instant-cli query '{ posts: {} }' --admin to query your app. A context flag is required: --admin, --as-email <email>, or --as-guest. Also supports --app <id>.
The bullets below are links to the Instant documentation. They provide detailed information on how to use different features of InstantDB. Each line follows the pattern of
Fetch the URL for a topic to learn more about it.
Think before you answer. Make sure your code passes typechecks tsc --noEmit and works as expected.
Remember! AESTHETICS ARE VERY IMPORTANT. All apps should LOOK AMAZING and have GREAT FUNCTIONALITY!
development
Maintainer-only workflow for handling GitHub Secret Scanning alerts on OpenClaw. Use when Codex needs to triage, redact, clean up, and resolve secret leakage found in issue comments, issue bodies, PR comments, or other GitHub content.
development
Maintainer workflow for OpenClaw releases, prereleases, changelog release notes, and publish validation. Use when Codex needs to prepare or verify stable or beta release steps, align version naming, assemble release notes, check release auth requirements, or validate publish-time commands and artifacts.
development
Run, watch, debug, and extend OpenClaw QA testing with qa-lab and qa-channel. Use when Codex needs to execute the repo-backed QA suite, inspect live QA artifacts, debug failing scenarios, add new QA scenarios, or explain the OpenClaw QA workflow. Prefer the live OpenAI lane with regular openai/gpt-5.4 in fast mode; do not use gpt-5.4-pro or gpt-5.4-mini unless the user explicitly overrides that policy.
development
End-to-end Parallels smoke, upgrade, and rerun workflow for OpenClaw across macOS, Windows, and Linux guests. Use when Codex needs to run, rerun, debug, or interpret VM-based install, onboarding, gateway smoke tests, latest-release-to-main upgrade checks, fresh snapshot retests, or optional Discord roundtrip verification under Parallels.