Developer docs

Public MCP connector

Add India Energy Atlas to Claude with one URL for read-only access to public data on India's electricity grid, with no sign-in.

Overview

A hosted Model Context Protocol server with 19 read-only tools for public data on India's electricity system: state demand (hourly and every 15 minutes), Power Supply Position, grid frequency, fuel mix, carbon intensity, demand forecasts, IEX prices, open-access charges and early-stage Data Centre site screens.

Setting	Value
Server URL	`https://api.energymap.in/mcp/public`
Transport	Streamable HTTP
Authentication	None. No account, key or sign-in.
Access	Read-only. No tool creates, changes or deletes anything.
Tools	19
Publisher, Claude connector and plugin	India Energy Atlas Pvt Ltd
Publisher, ChatGPT and Codex plugin	CIFR Frontier Research Council
Platform and data services	India Energy Atlas Pvt Ltd

Add it to Claude

Claude on the web and desktop

Open Settings, then Connectors, and select Add custom connector.
Name it India Energy Atlas and set the URL to https://api.energymap.in/mcp/public.
Leave authentication empty and save.
In a new chat, turn the connector on from the tools menu.

Claude Code

bash

claude mcp add --transport http india-energy-atlas https://api.energymap.in/mcp/public

India Energy Atlas plugin for Claude

The plugin adds this connector together with skills that pick the right tool and keep coverage and modelled-data labels in the answer. It contains no local code.

India Energy Atlas plugin on GitHub (opens in a new tab)

ChatGPT, Codex and other MCP clients

The same URL works in any MCP client that supports a remote Streamable HTTP server with no authentication, including the India Energy Atlas plugin for ChatGPT and Codex. Availability in a host's public directory depends on that host's review and publication process. Anthropic and OpenAI each handle your conversation under their own terms and privacy notices.

Tools

The server returns the current titles, descriptions and input schemas through tools/list. This table is a summary. Every tool is marked read-only and non-destructive. Only india_grid_data_centre_search is marked open-world, because it sends the place text to Mapbox.

Every state argument accepts a state slug or a display name, in any case.

Tool	What it returns	Key arguments
`india_grid_states`List covered Indian states	The states and union territories the Atlas covers, with state slugs, regions and headline metadata.	None
`india_grid_state_detail`Get one state energy profile	One state's capacity mix, peak demand, transmission coverage, build status and source records.	state_slug (required)
`india_grid_state_demand_15min`Get one state's 15-minute demand	The last 1 to 48 hours of closed 15-minute blocks in MW, each marked measured (the state's dispatch centre meter or Vidyut PRAVAH) or estimated (an Atlas fill until the real reading arrives), with coverage counts. An estimated block is not a measurement.	state (required), hours (1 to 48, default 24)
`india_grid_catalog`List or search the Atlas data catalogue	Atlas datasets with coverage, freshness and origin for each: every dataset, or those that match a search term.	query (optional search text)
`india_grid_data_centre_search`Find an Indian Data Centre location	Up to five India-only place matches with coordinates. Give a place, city or locality name, not a street address. The place text is sent to Mapbox for the lookup.	query (required, up to 120 characters)
`india_grid_data_centre_site_screen`Screen a Data Centre location in India	An early-stage development score from 0 to 100 with confidence, component scores for grid, demand, digital connectivity, water, industrial tariff and ecosystem, an indicative industrial energy-charge range and nearby infrastructure. Points outside India's land boundary are rejected.	latitude (required, 6.5 to 35.7), longitude (required, 68 to 97.5), label
`india_grid_data_centre_market_bands`Compare Indian Data Centre market bands	Named city or state screening bands on a 1 to 5 scale for one metric. A band of 0 means not reported.	metric (potential, demand, connectivity, water or industrial-tariff), area, minimum_band, limit (1 to 50)
`india_grid_live_snapshot`Get the live India grid snapshot	The latest, or a specified historical, grid snapshot across states: demand and frequency only, by state, with the IST time slot and coverage. Coverage counts official readings only.	ist (ISO timestamp in IST), coverage_threshold (0 to 1)
`india_grid_demand_accuracy`Score the modelled state-demand estimate	MAPE and MAE for the Atlas modelled state-demand estimate over a lookback window, for one state or pooled across states (`pooled_states`). It does not score the demand forecast.	days (1 to 90, default 14), state
`india_grid_demand_reconciliation`Get stored modelled-versus-official demand comparison	The stored comparison of Atlas modelled demand with official demand. Each row carries `computed_at`, `age_days` and `stale`.	window (24h, 7d or 30d), state
`india_grid_demand_forecast`Get an India demand forecast	An hourly electricity-demand forecast for India or one state. Every value is modelled.	horizon_hours (1 to 168, default 24), state
`india_grid_iex_heatmap`Get the IEX price heatmap	An hour-by-day view of recent IEX day-ahead market prices.	days (1 to 365, default 30)
`india_grid_iex_price_duration`Get the IEX price-duration curve	The share of recent IEX day-ahead hours above each price level.	days (1 to 1,095, default 30)
`india_grid_iex_area_prices`Get IEX area clearing prices	IEX area clearing prices by bidding zone, for one day or a lookback window.	date (YYYY-MM-DD), days (1 to 31, default 7)
`india_grid_frequency`Get recent grid frequency	Recent grid frequency in hertz, in up to 96 time buckets, each with mean, minimum and maximum. Values are the mean of state SLDC readings unless a row is marked as the NLDC national series.	hours (1 to 720, default 24)
`india_grid_posoco_psp`Get Power Supply Position data	Grid-India daily Power Supply Position: energy met, shortage, peak and outage by state or region.	days (optional, 1 to 90; omit for the latest day), state, region, include_state_rows
`india_grid_carbon_intensity`Get grid carbon intensity	Hourly modelled carbon intensity in gCO₂/kWh for a state or zone, with the modelling basis.	state, zone (northern, western, southern, eastern or north-eastern), hours (1 to 720, default 48)
`india_grid_carbon_mix`Get the fuel mix behind carbon	Generation shares by fuel (coal, gas, hydro, nuclear, wind and solar) behind the carbon estimates, for India or one state.	state, hours (1 to 168, default 24)
`india_grid_open_access`Get open-access electricity charges	Published open-access charges by state or DISCOM: cross-subsidy, additional, transmission, wheeling and banking charges, with effective dates.	state

Example prompts

Compare the latest electricity demand across Indian states. Cite the IST slot, coverage and source timestamps, and name the states with no data.
Find Navi Mumbai and screen its early-stage Data Centre development potential. Include confidence, missing inputs, industrial tariff caveats and what the score does not prove.
Give me the next 24 hours of electricity-demand forecasts for Karnataka and say clearly that they are modelled, not measured.
Show the southern region's grid carbon intensity over the last 24 hours and explain the units and modelling basis.
Give me India Energy Atlas's state profile for Delhi, including its sources and caveats.

Reading the results

Results keep the timestamp, unit and coverage fields that the underlying data provides. Missing states or values are labelled as missing.
Long series are summarised before they are returned. The aggregation field says how.
If the data service fails, a tool may return a recent cached copy. The result then carries served_from_cache.
Observed and modelled values are labelled separately. Demand forecasts, the modelled state-demand estimate and carbon intensity are modelled estimates, not measurements, and so is any 15-minute demand block marked as an estimate.
Data Centre screens are early-stage comparisons. They do not prove substation headroom, interconnection approval or dates, a tariff bill, water allocation, land title, permits or engineering feasibility.

Data handling

When Claude calls a tool, the server receives the tool name and its arguments, such as a state slug, date, lookback period, place name or coordinate. It does not receive your conversation, and it never asks for your name, email address, password, payment details or an API key. Security logs are kept for up to 90 days. For the Data Centre place search, the place text is sent to Mapbox, Inc. without your IP address or any identifier.

Limits

20 requests per second per IP address.
The Data Centre tools share a budget of 300 calls per 10 minutes across all clients.
Public data refreshes every 15 minutes to daily, depending on the dataset.
When a limit is reached, or the data service is slow and no cached copy is available, the tool returns a short message asking you to try again shortly.

Support

Email hello@energymap.in. India Energy Atlas Pvt Ltd operates the Atlas platform and data services and publishes the Claude connector and plugin. CIFR Frontier Research Council publishes the plugin for ChatGPT and Codex.