startup-india-guide

Every central government scheme for Indian startups — one searchable, comparable, verified documentation site.

The Government of India's 107-page Playbook of Government Schemes and Initiatives for Startups (June 2026), rebuilt as a docs site: 69 schemes with verbatim eligibility, benefits and official links, organised the way a founder actually looks for money — by need, by stage, by sector.

Live site · Explore India map · Scheme Finder · All 69 schemes · Compare

The home page. Palette sampled from the source PDF's own navy, saffron and green.

What it looks like

Four views from the live site — dark home, the filterable directory, a scheme one-pager, the finder's results. Light/dark theme, ⌘K search, and mobile-first layout throughout.

What's on the site

Page	What it does
Overview	Stats, browse by support / stage / sector, flagship schemes
Scheme Finder	The playbook's official 5-question decision tree as an interactive wizard
All schemes	Filter 69 schemes by part, support type, stage, sector, audience — cards or table, deep-linkable (?support=grant&stage=ideation)
Scheme pages	One per scheme: plain-English summary, eligibility checklist, benefits, how to apply, verified official links, related schemes, source-page citation
Compare	Up to three schemes side by side
Lifecycle Map	Ideation → prototype → seed → growth → market access
What do you need?	From a need (grant / loan / lab / buyers / IP) to the schemes that provide it
Explore India map	Choose a full-screen 3D walkable India experience, the conventional incubator map, or the state-schemes map. The walkable view includes keyboard/touch movement, drag-to-look, pinch zoom, state wayfinders, type-coloured incubator pins, and policy-rich state drawers
State & UT schemes	320+ state-level startup schemes & incentives (seed grants, subsidies, reimbursements, procurement) across every state and UT — straight from each official state startup policy, by-state / all-schemes / map views. Kept separate from the central schemes
Incubators directory	220+ technology business incubators, Atal Incubation Centres & startup hubs on an interactive India map (choropleth + city markers) — searchable, filterable, cards / table / state-wise views, with locations, websites and contacts
PSU & regulators · States & UTs	17 PSU/regulator programs; every state startup portal
Glossary · About	Definitions, abbreviations, disclaimer, method

How it works

Startup-Schemes-Playbook-June-2026.pdf          the source document
      │
      │  extract (text + page images) → parse → adversarially verify
      ▼
data/*.json                                     the content: 69 schemes, decision tree,
      │                                         lifecycle, needs index, PSU/states, glossary
      +
site/tokens.css + site.css + site.js            the skin (PDF-sampled palette) + engine
      │
      │  node build.mjs                         zero dependencies, fails loudly
      ▼
docs/  →  GitHub Pages                          85 static pages, ⌘K search, no runtime deps

1. Extract — every PDF page rendered to text and image; scheme one-pagers parsed to structured JSON; hyperlinks taken from the PDF's link annotations, never retyped.
2. Verify — an independent pass re-reads each source page and corrects amounts, bullets and URLs before anything ships.
3. Build — build.mjs turns data/ + site/ into docs/: directory, wizard, compare, search index, sitemap, per-scheme pages.
4. Deploy — pushing to main publishes docs/ to the gh-pages branch via the included workflow.

Use this as a template

The engine is content-agnostic: it turns any structured reference corpus — a scheme catalogue, a benefits handbook, an internal policy binder — into this site. Click Use this template, then:

1. Point it at your home. Edit site.config.json — siteName, siteBase, pathPrefix, repo.
2. Bring your content. Replace data/*.json (schemes, decision tree, lifecycle, needs index, glossary). The shapes are plain and documented by example; data/aliases.json resolves any cross-reference spellings.
3. Wear your brand. site/tokens.css is the only file that knows the palette — light and dark themes, ~60 custom properties.
4. Build & publish. node build.mjs, push, and enable Pages (the included workflow publishes docs/ → gh-pages).

The build fails loudly on duplicate slugs, missing fields, malformed URLs, or any scheme name that doesn't resolve to a page — broken cross-references can't ship silently.

Repo tour

startup-india-guide/
├── Startup-Schemes-Playbook-June-2026.pdf   # the source document (also served by the site)
├── site.config.json         # template knobs: site name, base URL, path prefix, repo
├── data/                    # the extracted, verified content — single source of truth
│   ├── schemes.json         # 69 schemes: verbatim eligibility/benefits/links + tags
│   ├── decision-tree.json   # the 5-question finder
│   ├── lifecycle.json       # stage → schemes map
│   ├── needs-index.json     # need → schemes map
│   ├── state-schemes.json   # 320+ state/UT startup schemes & incentives, by state
│   ├── incubators.json      # 220+ incubators: location, type, sectors, website, contacts, lat/lng
│   ├── india-map.json       # projected + simplified India state/UT outlines (see scripts/) for the maps
│   ├── india-terrain.json   # relief, rivers, lakes and named peaks for the walkable map
│   ├── psu.json · states.json · glossary.json · about.json
│   └── aliases.json         # printed-name → slug overrides for cross-references
├── site/                    # the engine — consumed by build.mjs
│   ├── tokens.css           # the skin: PDF-sampled palette, light + dark
│   ├── site.css             # layout & component vocabulary; reads only tokens
│   ├── site.js              # search, filters, wizard, compare, incubator map, theme toggle
│   ├── walkable-core.js     # pure movement, zoom, timing and placement helpers
│   ├── walkable-3d-core.js  # pure terrain/height-field, noise and camera-rig math
│   ├── walkable-3d.js       # full-screen 3D India world, controls, wayfinders and state drawers
│   └── vendor/              # three.js module build (MIT), the map's only runtime dependency
├── build.mjs                # data + site → docs/   (zero dependencies, Node ≥ 18)
├── WALKABLE_MAP.md          # walkable-map product and implementation contract
├── TODOS.md                 # future product work and acceptance criteria
├── scripts/geo.mjs               # projection/simplification helpers shared by the two data-prep tools
├── scripts/build-india-map.mjs   # one-off: India GeoJSON → data/india-map.json (state outlines + projection)
├── scripts/build-india-terrain.mjs # one-off: Natural Earth → data/india-terrain.json (same projection)
├── docs/                    # GENERATED — never hand-edit; what GitHub Pages serves
└── .github/workflows/deploy-pages.yml   # publishes docs/ → gh-pages on push to main

Working on the site

npm run build                         # regenerate docs/ from data/ + site/
npm test                              # walkable-map layout, movement, zoom and timing checks
npm run dev                           # build and preview at http://localhost:8000

• Content change (an amount, a new link, a new scheme)? Edit data/*.json, rebuild.
• Design change? Edit site/tokens.css (palette/fonts) or site/site.css (components), rebuild.
• Never edit docs/ — every build overwrites it.

Roadmap

The walkable map is single-player today. A future live-explorers feature would give each visitor an anonymous local avatar, show an approximate live explorer count, and render nearby visitors through an external realtime presence service. The MVP is intentionally ephemeral and privacy-minimal: no accounts, chat, GPS, permanent profiles, or location history.

How the data was made

Each of the playbook's 107 pages was extracted to text and rendered to an image; every scheme one-pager was parsed into structured JSON and then independently re-verified against the source page — ₹ amounts, eligibility bullets, benefits and URLs are reproduced as printed, and each scheme page cites the playbook page it came from. Hyperlinks come from the PDF's link annotations, never reconstructed by hand.

Honest limitations

• Scheme details drift. Deadlines, corpus sizes and application windows change after any snapshot; every page carries a "verify on the official portal" notice for exactly that reason.
• This is an independent reference, not a government website, and not legal or financial advice.
• Extraction is machine-verified, not lawyer-reviewed. The verification pass catches transcription errors; it cannot catch changes the government made after June 2026.

Map data

The maps are built from two committed, regenerable datasets. Both are produced by the one-off tools in scripts/, share a single projection descriptor, and are never fetched at run time.

Dataset	Source	Terms
data/india-map.json — state/UT outlines	udit-001/india-maps-data	MIT
data/india-terrain.json — relief, rivers, lakes, peaks	Natural Earth 10m physical, via martynafford/natural-earth-geojson	Public domain

Natural Earth asks for no attribution; it is credited here anyway. Terrain names are transliterated to their conventional Indian spellings — the renames are listed explicitly in scripts/build-india-terrain.mjs. A relief region is only named on the map when a meaningful share of it lies inside India, so features that merely graze the border are drawn but left anonymous.

The walkable map renders with three.js (MIT), vendored in site/vendor/ so the site keeps building with zero npm dependencies.

License

MIT — see LICENSE. The underlying playbook is a Government of India publication; this repo reproduces its content as a reference with attribution and a link to the source PDF.