India's DPDP Act, built into your Node.js backend.
Declare why you hold each piece of personal data in one typed file. dpdp-kit keeps consent records you can prove, carries out erasure on time, tracks rights requests, and fails CI the moment your schema drifts.
The penalties are real money.
The Act's Schedule (Section 33) sets the maximum penalty for each kind of breach. Even the line for "any other provision", which covers consent, notices, erasure and rights requests, goes up to ₹50 crore.
The burden of proof is on you.
If consent is questioned, Section 6(10) makes the Data Fiduciary prove that a notice was given and consent was given. "consented: true" in a column isn't proof.
May 2027 is closer than it looks.
Notices, erasure, breach handling and rights requests all start then. Retrofitting them into a grown codebase takes longer than building them in now.
Spreadsheets drift.
A data map that lives outside the code is out of date the day someone adds a column. Compliance has to live where the changes happen: in the code and in CI.
Someone adds a column. CI notices.
Most teams keep their data map in a spreadsheet, and six months later it's wrong. dpdp-kit reads your real Drizzle, Prisma or Mongoose schema and fails the build when a personal-looking column has no declared purpose, or when the map points at columns that are gone.
Every duty, mapped to code.
Each Section and Rule below was checked against the official Gazette texts of the Act, the Rules and their corrigendum.
| The law says | Where | dpdp-kit |
|---|---|---|
| Process data only for a stated purpose, with consent or a legitimate use | Sections 4, 6, 7 | A typed data map: every field lists its purposes |
| Prove that a notice was given and consent was given | Section 6(10) | Hash-chained consent records with the notice version and language |
| A notice in English or any Eighth Schedule language | Sections 5(3), 6(3) | All 22 languages built in |
| Erase on withdrawal or when the purpose ends, unless another law says keep it | Section 8(7) | Erasure plans and a retention runner, with legal holds |
| Keep data, traffic data and logs of processing for at least a year | Rule 8(3) | minimumRetention, enforced by the audit |
| Log, monitor and review who accesses personal data, and keep the logs for a year | Rule 6(1)(c), (e) | An access log with review, flags reads outside a field's purpose |
| Encryption, masking or virtual tokens | Rule 6(1)(a) | dpdp-kit/protect: AES-256-GCM, keyed tokens, masking |
| Tell people and the Board about a breach, with a detailed report within 72 hours | Section 8(6), Rule 7 | Breach records with deadlines, notices and the Board report |
| Answer access, correction, erasure, grievance and nomination requests | Sections 11–14 | Rights requests with due dates |
| Answer grievances within 90 days | Rule 14(3) | A default due date, enforced by the audit |
| Tell processors to stop, and to erase | Sections 6(6), 8(7)(b) | Processor hooks, or a manual task when there's no hook |
One config. Five things it does for you.
Say why you hold each field.
Purposes, their legal basis and how long you keep the data live in one file. Purposes are typed, so a misspelt one is an error in your editor, and the audit catches the rest.
The config, field by field →// dpdp.config.ts
import { defineConfig } from "dpdp-kit";
export default defineConfig({
fiduciary: { name: "Acme Pvt Ltd", grievanceUrl: "https://acme.in/grievance" },
purposes: {
account: { basis: "legitimate-use", retain: "while-active" },
marketing_sms: { basis: "consent", retain: { afterWithdrawal: "0d" } },
},
data: {
users: {
table: "users",
subject: "id",
fields: { phone: { category: "contact", purposes: ["account", "marketing_sms"] } },
},
},
minimumRetention: "1y", // Rule 8(3)
});Consent you can prove.
Section 6(10) puts the burden of proof on you. Every grant and withdrawal is stored with the notice version, language and channel, in a per-person SHA-256 hash chain.
verify() catches an edited, deleted or reordered event.
await dpdp.consent.grant(userId, ["marketing_sms"], {
noticeVersion: "2026-11",
language: "hi", // English or any of the 22 Eighth Schedule languages
channel: "android",
});
await dpdp.consent.has(userId, "marketing_sms"); // true
await dpdp.consent.verify(userId); // { ok: true, events: 1 }Erasure that actually happens.
- Processors are told to stop straight away.
- A field is erased only when none of its purposes remain.
- Rule 8(3)'s one-year minimum is respected.
- Legal holds win: GST invoices wait 8 years, then get anonymised.
- Safe with several servers: tasks are claimed atomically.
// a withdrawal schedules the erasure as dated tasks
await dpdp.consent.withdraw(userId, ["marketing_sms"]);
// an account deletion, with legal holds applied
await dpdp.erasure.deleteAccount(userId);
// then, from a daily cron:
const report = await dpdp.retention.run(); // { done, failed, manual }Rights requests, on a clock.
Access, correction, erasure, grievance and nomination requests (Sections 11–14) are recorded with a due date. Grievances default to the 90 days of Rule 14(3).
Rights requests →const request = await dpdp.rights.open(userId, "grievance", {
text: "I keep getting promo SMS",
}); // due in 90 days
await dpdp.rights.list({ overdue: true }); // your inbox of late answers
await dpdp.rights.summary(userId); // Section 11: what you hold and whyA GitHub Action that comments where it hurts.
Each problem lands on the pull request as an annotation, with a summary on the run page and errors / warnings outputs.
# .github/workflows/dpdp.yml
on: [pull_request]
jobs:
audit:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: actions/setup-node@v4
with: { node-version: 22 }
- run: npm ci
- uses: yogesh2902/dpdp-kit@v0
with:
schema: src/db/schema.tsCompliance that lives in your code.
There are three common ways to handle the DPDP Act. Only one of them is checked every time your code changes.
| Spreadsheet and a policy page | Hosted consent platform | dpdp-kit | |
|---|---|---|---|
| Checked on every pull request | No | Usually not | Yes, in CI |
| Catches a new personal column before it ships | No | Usually not | Yes, from your schema |
| Personal data stays on your servers | Yes | Usually sent to the vendor | Yes, no network calls |
| Consent records that show tampering | No | Varies | Yes, hash-chained |
| Erasure carried out on time, with legal holds | By hand | Varies | Yes, retention runner |
| Mistakes caught in your editor | No | No | Yes, typed config |
| Cost | Free, but manual | Subscription | Free, MIT |
Compliance as code
One typed config is the single source of truth. The audit, the guards, erasure and rights requests all read it.
Drift-proof
The audit reads your real Drizzle, Prisma or Mongoose schema in both directions, so the data map can't quietly fall out of date.
Node-native, no SaaS
A library inside your process. No server to run, no account, no API key, and your own Postgres or MongoDB.
Checked against the law
Every Section and Rule reference was checked against the official Gazette texts. A wrong reference is treated as a bug.
Tested like infrastructure
250+ tests, on real Postgres in IST, Prisma and MongoDB, with runners racing for the same tasks. Example apps go through a year of customers on every change.
Open, no lock-in
MIT licensed. Your records live in your own tables, and every store and data-access layer is an interface you can replace.
Any team holding Indian users' data.
If your app stores a phone number, an email or an address of someone in India, the Act applies. Here's where dpdp-kit helps most.
Fintech and lending
KYC records under legal holds, erasure that respects them, and rights requests with due dates.
E-commerce and D2C
Consent for marketing SMS and email, GST invoices kept for 8 years then anonymised, and SMS vendors told to stop.
Ed-tech
Consent records today. Children's data and verifiable guardian consent (Section 9, Rule 10) are on the roadmap.
Health and wellness
Health and biometric columns are flagged by the drift check, so nothing is stored without a stated purpose.
SaaS and B2B
Processor hooks tell analytics and messaging vendors to stop and erase, and the Action reviews every pull request.
Agencies and freelancers
One config and one CI check per client project, so every app you ship starts with the basics in place.
The clock is already running.
The DPDP Rules, 2025 were published on 13 November 2025 and start in phases (Rule 1). Building this in now leaves time to test.
Not legal advice. dpdp-kit helps you put compliance into practice; it doesn't make an app compliant on its own. Section and Rule references point you to the text so you can check them. Have your own lawyer review your purposes, notices and retention periods.
What's shipped, and what's next.
dpdp-kit is growing from a backend toolkit into a full-stack one: the same config will drive the screens people see, on the web and on mobile.
- Typed config and audit CLIpurposes, data map, checks
- Schema drift checkDrizzle, Prisma and Mongoose
- Your databaseDrizzle, Prisma, MongoDB and Mongoose, or any Postgres client
- Breach recordsSection 8(6), Rule 7: who to tell, and the Board's 72-hour report
- Consent recordshash-chained, 22 languages + English
- Erasure and retention runnerlegal holds, Rule 8(3), NOT NULL replacements
- Rights requestsSections 11–14, Rule 14 due dates
- Security safeguardsRule 6: access log, encryption, tokens, masking
- Guards for every frameworkExpress, Hono, Fastify, NestJS
- Events and daily upkeep
dpdp.on,dpdp.daily() - GitHub Actionannotations on pull requests
- Example appsPrisma and Mongoose, a year of customers in CI
- Rule 3 notices from the configin every offered language, with an automatic version
- Objections to a legitimate useSection 7(a)
- HTTP APIExpress, Hono and Next.js adapters
- Rights requests and transferscorrections applied, cross-border records (Section 16)
- MySQL and SQLitedata access
- React and React Nativea consent notice and a privacy centre
- Children's dataage gate and verifiable guardian consent, Section 9, Rule 10
- Inactivity erasure48 hours' warning, Rule 8(1)–(2)
- Erasure beyond SQLfiles, caches, search indexes
- Consent Manager interopSections 6(7)–6(9), Rule 4
Plans can change as the law and the community shape them. Want something sooner? Open an issue or star the repo to follow along.
Things people ask.
Does dpdp-kit make my app compliant?
No tool can do that on its own. dpdp-kit puts the obligations into practice in your code: declared purposes, consent records, erasure and rights requests. Your company still writes its notices, signs its vendor contracts, runs access control and backups, and has its own lawyer review its purposes and retention periods. Who does what lists every duty in the Act and Rules, and which side does it.
Is it free?
Yes. It's MIT licensed, with no paid tier and no account. You can use it in commercial products.
Do I need to run a server or sign up for anything?
No. It's a library that runs inside your Node.js process and writes to your own database. There's no API key, and it makes no network calls.
Which databases and frameworks does it support?
Storage and data access work with Postgres through Drizzle, Prisma or any Postgres client (node-postgres, postgres.js, Kysely, Knex), and with MongoDB and Mongoose. The schema drift check reads Drizzle schemas (Postgres, MySQL, SQLite), Prisma schemas and Mongoose models. Guards are included for Express, Hono, Fastify and NestJS. Every store is an interface, so you can add another database. MySQL and SQLite data access come in 0.6.
Which Node.js versions?
Node.js 20.19 or later. The package is ESM-only, and require("dpdp-kit") also works on those versions.
When do I need to have this in place?
The DPDP Rules were notified on 13 November 2025. Most obligations, including notices, erasure, breaches, children and rights requests, start in May 2027 (Rule 1). Starting now leaves time to test.
What if a legal reference is wrong?
Every reference was checked against the official texts of the Act, the Rules and their corrigendum, but laws change. Report it as a GitHub issue; it's treated as a bug.
How can I help?
Star the repo, try it in a project, and open issues for what's missing. Pull requests are welcome, especially for new database adapters.