Every claim links to its official government source. Tap any source to verify.

NirmanKit

Methodology

Where this content comes from.

What we count as a source, how we verify, how often we re-check, and how our data lives in India.

Last updated · 23 May 2026

What counts as a source

A source must satisfy four tests, per our editorial guide:

  1. Government-issued — ministry portal, PIB release, Gazette notification, Parliamentary answer (Lok Sabha or Rajya Sabha), Budget speech annexure, RBI / SEBI circular, ministry FAQ. Reputable secondary sources (PRS, IBEF, FICCI) support but do not substitute primary sources.
  2. Public — accessible without paywall, login, or pay-per-page.
  3. Dated — has a publication date, or where missing, an explicit archive date when we fetched it.
  4. Stable — has a canonical URL. Where the URL routinely shifts, we also store an archive snapshot.

How we verify

Every scheme is re-verified at least every 90 days, by a named editor, against the cited sources. The editor confirms:

Where a scheme is unchanged, re-verification is a 30-second confirmation and the last_verified_at timestamp advances. Where it has changed, we open a content PR and a second editor reviews before publish.

A daily automated job watches the canonical URLs for content drift; when a source page changes, it raises an alert into the editorial queue so the editor knows to revisit before the 90-day deadline.

Our eligibility rules are open

Every rule that decides "are you eligible for X?" is openly licensed under ODbL 1.0. In plain language: anyone may copy, redistribute, and build on the rules dataset — even commercially — as long as they credit NirmanKit and share any adapted dataset under the same licence.

You don't have to download anything to audit us. Every rule that decides whether you're eligible is shown on the scheme page itself, with the rule version and effective-from date. How the rules are written, where they run, and what they're not allowed to do is summarised in our eligibility-engine policy summary.

Researchers, journalists, and partners who want the full dataset can request a copy via the contact page.

Scheme content (the prose) is licensed CC-BY-4.0 — fork, translate, republish; please cite us.

Where the data lives

We follow an India-region default for all personal data. Specifically:

The one cross-border surface is the AI-assisted DPIIT drafter (Phase 2 / V1.1), which calls the Anthropic API outside India. That's opt-in, no PII flows in the prompt, and we benchmark Indian providers (Sarvam, AI4Bharat, Bhashini) every six months with a published switch trigger.

Full details: our security and privacy posture and where your data is stored.

Every source we currently cite

For the 155 seeded schemes, these are every canonical URL we cite. We re-fetch them on the cadence above.

If we got something wrong

Use the contact page to report any inaccuracy. We triage within one business day and fix or respond within 7 calendar days — that's a public commitment, not a hope.