Session Management

Blitz has built in session management that can be used with any type of authentication or identity providers.

Session management performs the following functions:

  1. Tracking whether a user is logged in or not
  2. Attribute multiple requests to the same user, even when they are logged out
  3. Protection against CSRF attacks


You login, logout, and otherwise modify the session via the

SessionContext object which is accessible anywhere on the server.


For login you will have form component in your UI that will submit a login mutation like the one shown below.

// app/auth/mutations/login.ts
import {Ctx} from "blitz"
export default async function login(input: SomeTSInputType, ctx: Ctx) {
// 1. Validate input data
// 2. Validate user credentials
// 3. Fetch user data
// 4. Create a new session (log in)
await ctx.session.create({userId:, roles: [user.role]})


For logout you will have a button in your UI that will submit a logout mutation like the one shown below.

Revoking a session will immediately delete all client-side query cache causing all queries on the page to be refetched. This ensures any sensitive data in the cache is deleted.

// app/auth/muations/logout.ts
import {Ctx} from "blitz"
export default async function logout(_: any, ctx: Ctx) {
// 1. Revoke the current user session, logging them out.
return await ctx.session.revoke()

Change Session PublicData of Current User

Each session has

PublicData, data which is available on the client and has the potential to be read by third-parties because it's stored in a cookie readable by any Javascript code. This is usually used to store the current user id, user role, and perhaps current organization id.

You can change the session public data in any query or mutation like this:

// app/mutations/someMutation.ts
import {Ctx} from "blitz"
export default async function someMutation(input: any, ctx: Ctx) {
// This merges the input data with whatever is already in current publicData
await ctx.session.setPublicData({orgId: 1})

Access Session on the Server

In Queries & Mutations

SessionContext is available off of ctx which is provided as the second parameter to all queries and mutations because of the sessionMiddleware that's in blitz.config.js.

// app/queries/someQuery.ts
import {Ctx} from "blitz"
export default async function someQuery(input: any, ctx: Ctx) {
// Access the SessionContext class

In getServerSideProps or API Routes

You can also get the session context inside

getServerSideProps or inside API routes with getSessionContext like this:

import {getSessionContext} from "@blitzjs/server"
export const getServerSideProps = async ({req, res}) => {
const session = await getSessionContext(req, res)
console.log("User ID:", session.userId)
return {props: {}}

Access Session on the Client

Blitz provides a

useSession() hook that returns PublicData with isLoading property. This hook can be used anywhere in your application.

import {useSession} from "blitz"
function SomeComponent() {
const session = useSession()
return /*... */

Production Deployment Requirements

In production, you must provide the

SESSION_SECRET_KEY environment variable with at least 32 characters. This is your private key for signing JWT tokens.

Anonymous Sessions

If a user is not logged in, an anonymous session will automatically be created for them. You can use

ctx.session.setPublicData() and ctx.session.setPrivateData() for anonymous sessions the same as for logged in users. Any data you set for an anonymous session will automatically be transferred to an authentication session when a user logs in.

Anonymous sessions are JWT tokens that are stored on the client as an httpOnly cookie that never expires.

PublicData for anonymous sessions is kept in the session JWT and not stored in the database. Anonymous sessions will only be saved in your database if you call session.setPrivateData().

The anonymous session will be created on the first network request, whether SSR or via an API. This will happen as long as

sessionMiddleware is in your middleware chain for that request.

One use case for this is saving shopping cart items for anonymous users. If an anonymous user later signs up or logs in, the anonymous session data can be merged into their new authenticated session.

Anonymous session

PublicData looks like this:

userId: null,
roles: []

Customize Session Public Data in Typescript

If using Typescript, first update the

PublicData type in types.ts like this:

import {DefaultPublicData} from "blitz"
import {User} from "db"
declare module "blitz" {
export interface PublicData extends DefaultPublicData {
userId: User["id"]
roles: readonly ["admin" | "customer"]
orgId: number

Then change all uses of

ctx.session.create() to pass in the new fields.

ctx.session.create({userId: 1, roles: ["admin"], orgId: 1})

You can also use

ctx.session.setPublicData() to update session data for an already logged in user. This will merge values with the existing public data.

ctx.session.setPublicData({orgId: 1})

Session Configuration

You can customize session management by passing an object to the

sessionMiddleware factory function.

// blitz.config.js
const {sessionMiddleware, simpleRolesIsAuthorized} = require("@blitzjs/server")
module.exports = {
middleware: [
sessionExpiryMinutes: 1234,
isAuthorized: simpleRolesIsAuthorized,

Available options:

type SessionConfig = {
sessionExpiryMinutes?: number /* Default: 30 days */
sameSite?: "strict" | "lax" | "none" /* Default: 'lax' */
domain?: string /* Default: undefined. Can set as `` to work across subdomains */
getSession: (handle: string) => Promise<SessionModel | null>
getSessions: (userId: string | number) => Promise<SessionModel[]>
createSession: (session: SessionModel) => Promise<SessionModel>
updateSession: (handle: string, session: Partial<SessionModel>) => Promise<SessionModel>
deleteSession: (handle: string) => Promise<SessionModel>
isAuthorized: (userRoles: string[], input: any) => boolean
interface SessionModel extends Record<any, any> {
handle: string
userId?: string | number
expiresAt?: Date
hashedSessionToken?: string
antiCSRFToken?: string
publicData?: string
privateData?: string

Customize Session Persistence & Database Access

By default, session persistence is zero-config with Prisma. However, you can customize this to save sessions somewhere else, like Redis. You can also customize this if you have Prisma but want to customize the attribute names on the user or session model.

Customize session persistence by overriding the database access functions defined above in

SessionConfig. The functions can do anything, but they must conform to the defined input and outputs types.

For reference, here's

the default config that works with Prisma.

Manual API Requests

When making a request from the client to an API route, you need to include the anti-CSRF token in the

anti-csrf header like this:

import {getAntiCSRFToken} from "blitz"
const antiCSRFToken = getAntiCSRFToken()
if (antiCSRFToken) {
// Set fetch request header["anti-csrf"] = antiCSRFToken

And then you can get the sessionContext in the API route like this:

import {getSessionContext} from "@blitzjs/server"
export default async function ({req, res}) {
const session = await getSessionContext(req, res)
console.log("User ID:", session.userId)

Technical Details of How it Works

Authenticated sessions use opaque tokens that are stored in the database.

Implementation Details

Session Creation

  • At login, the server creates two opaque tokens:
    • An access token.
    • An anti-csrf token.
  • Both are a 32 character long string.
  • The access token is sent to the frontend via an httpOnly, secure cookie.
  • The anti-csrf token is sent to the frontend via a normal, secure cookie that can be read from Javascript.
  • The SHA256 hash of the access token will be stored in the database. This token has the following properties mapped to it:
    • userId
    • expiry time
    • session data
  • The anti-csrf token is stored alongside the access token.
  • Creating a new session while another one exists results in the headers / cookies changing. However, the older session will still be alive.
  • For serious production apps, a cronjob is needed to remove all expired tokens on a regular basis.

Session Verification

  • For each request that requires CSRF protection, the frontend sends the anti-csrf token in the request header.
  • An incoming access token is verified by checking that it's in the db and that it has not expired. After each verification, the expiry time of the access token is updated.
  • CSRF attacks are prevented by checking that the incoming anti-csrf token (from the header) is the one associated with the session.

Session Revocation/Logout

  • This is done by deleting the session from the database.
  • Logout additionally clears the cookies, and a header is sent signaling the frontend to remove the anti-csrf token from the localstorage



interface SessionContext {
* null if anonymous
userId: unknown
roles: string[]
handle: string | null
publicData: PublicData
authorize: (roleOrRoles?: string | string[]) => void
isAuthorized: (roleOrRoles?: string | string[]) => boolean
create: (publicData: PublicData, privateData?: Record<any, any>) => Promise<void>
revoke: () => Promise<void>
revokeAll: () => Promise<void>
getPrivateData: () => Promise<Record<any, any>>
setPrivateData: (data: Record<any, any>) => Promise<void>
setPublicData: (data: Record<any, any>) => Promise<void>


By default,

PublicData is typed as

interface DefaultPublicData {
userId: any
roles: string[]

but you can, and should, extend

PublicData to match your application. You do this as shown below.

// types.ts
import {DefaultCtx, SessionContext, DefaultPublicData} from "blitz"
import {User} from "db"
declare module "blitz" {
export interface Ctx extends DefaultCtx {
session: SessionContext
export interface PublicData extends DefaultPublicData {
userId: User["id"] // strongly types userId
orgId: number // adds new field to PublicData

And then

PublicData everywhere in your application, including ctx.session and useSession will use this type that you define.

Idea for improving this page?Edit it on GitHub
Bytes Newsletter