Are they really useful?

Trees first, and every number checkable

The story is told in a fixed order of importance, and it has to hold up when a developer, an official or a neighbour checks it.

1. The forest. The trees and the whole Sahyadri ecosystem: species found nowhere else, corridors, sacred groves.
2. The process. How environmental reviews are sidestepped, and how projects are imposed on local and Adivasi communities.
3. The energy case. Power, emissions and cost, as supporting evidence — including what supporters say.

Decisions are still being made

Study terms, clearances, gram sabhas and public hearings are still ahead for most projects. The Take action page says what a reader can do, and who is already working on the ground.

Groups working on this

Home, and a story in ten chapters

Home opens on a Sahyadri photo and four headline numbers, then an animated explainer, How a hill becomes a battery: cheap daytime power, dear evening power, the cost to the forest, the cost to life around it. A fair box says what pumped storage is good for, and cards lead to every page.

Story is the long read: 48,557 trees · What lives here · The forest bill · How the review was sidestepped · Imposed, not designed · The promise · A battery that can burn coal · Built, not used · What supporters say · What I ask for. Each chapter ends with a share card.

Projects: 39 dots on one ridge

Every pumped storage project planned in Maharashtra, 60,615 MW in all, in one table and on one map.

• The table is sorted by trees to be felled, then forest land, then capacity — never by megawatts first. Filter by status, or show Western Ghats projects only.
• The map (Leaflet on OpenStreetMap) places each project from its coordinates; dot size is MW, colour is the clearance stage.
• A page per project: MW, trees, forest hectares, villages, a progress bar from MoU to operation, its timeline, its documents and a WhatsApp button.

Timeline and news, every morning

Timeline: 89 events — MoUs, study terms, clearances, felling orders, hearings and protests — newest first. It opens on the “Forest & people” filter, and can be narrowed by year, category, project or verified-only.

News: reports from a fixed list of 75 vetted outlets, newest first, with tag filters that put trees, ecosystem and consent first. A daily sync adds new items; anything not yet checked by a person carries a not yet reviewed badge. The same list is the /news.xml RSS feed.

Shortcuts, and the voices they skip

Shortcuts: 13 cards, each naming a safeguard in environmental or forest law, what happened instead, and the documents that show it. Example: the land law requires a resettlement plan, but Karjat’s draft EIA calls it “not applicable” by citing a 1,000 ha threshold — for a project that needs 377 ha. Filter by project; 7 are shown in official documents.

Voices: 14 records from the affected villages — gram sabha resolutions, public hearings and objections — attributed by name, with an evidence label on each.

How the numbers are worked out

Methodology: 11 inputs and every formula behind the model — forest per GWh stored, round-trip loss (100 units in, about 77.5 out), emissions on today’s grid, and battery vs pumped storage tariffs from recent tenders. It is labelled a model, not a measurement, and every input names its source.

Glossary: 44 terms in plain words — ToR, EC, FC, ESA, gram sabha — grouped by category, each with its own link.

About: the source policy, the full outlet list, corrections, the right of reply to developers, and the data downloads.

Git is the database; a build is the check

1. Storage — one JSON file per entity in data/: projects, timeline, news, shortcuts, consent, ecology, outlets, glossary, facts, citations. Every change is a commit.
2. Data layer — zod schemas are the one definition of every shape; loaders, cross-file integrity checks, selectors, the energy model and the { en, mr } text helpers sit on top.
3. Pages — Astro renders each page once per language into one static URL. A schema error fails the build, so bad data never ships.
4. Daily sync — a Claude scheduled task runs the sync-ghats-data skill at 07:40 IST: search whitelisted outlets, dedupe, add as unreviewed, validate, commit.
5. Hosting — Cloudflare Pages rebuilds on every push.