The seizures India's enforcement agencies announce, in one public index.

Seizure figures are public but scattered. They sit in prose inside thousands of ministry releases, in units such as lakh, crore, MT and quintal, and in valuations of "worth Rs 10 crore". PIB alone issues well over a thousand English releases a month. A reporter who wants to know how much gold was seized along the Bangladesh border, or which agency reports most drug seizures at sea, has to read them one by one.

Counts that do circulate are often totals from speeches or replies in Parliament, with no way to see the events behind them.

• Turns each seizure release into one row: category, quantity in a standard unit, stated value, district, state, agencies, release date and seizure date where given, and the source link.
• Separates land-border districts from coastal and inland districts, airports, ports and seizures at sea, and marks each event border-related or not, with the reason.
• Shows monthly series, state and district tables, agencies and the largest seizures, filterable by category.
• Offers a searchable event table and a CSV of any filtered view, plus a full CSV and JSON for re-use.
• Keeps a separate, labelled layer of seizures reported in Indian news, so official and reported figures are never mixed.
• Publishes its rules, its source status and its gaps on a methods page.

1. ListMonthly PIB listings: Delhi for every ministry, nine regional offices for Finance, Home Affairs and Defence.
2. ScreenTitles that name an action and contraband go on. Ceremonies, disposal drives and Parliament replies stop here.
3. ReadRules read quantities, units, values, agencies, places and dates from the title, subtitle and opening paragraphs.
4. Place and mergeDistrict and zone from a Wikidata gazetteer; releases about the same seizure become one event with every source.
5. PublishNames are stripped, every export is scanned for personal data, and the site is rebuilt from committed files.

Every response is cached, so the index can be rebuilt offline. Requests are spaced and identified. Details, formulas and unit conversions are on the methods page.

Compiled data is published under CC BY 4.0.

• Journalists checking an official claim, finding the release behind a number, or tracing a pattern across months and states.
• Researchers who need a clean, sourced event table instead of a stack of PDFs and web pages.
• Citizens and legislators asking what enforcement in their district looks like in the public record.
• Agencies that want their published record shown accurately, and a way to report a correction.

• Events, not people: no names of accused, witnesses or victims, no person counts, no profiles (doctrine red lines 1 and 3).
• No identity analytics: nothing is filtered or totalled by nationality, religion, caste, ethnicity or language. Nationality in a title becomes "foreign national" (red line 2).
• Every export is scanned for phone, Aadhaar-like, PAN, passport and vehicle numbers and honorific names before it is written; a finding stops the run.
• Every number links to its source and carries an as-of time. Official and news figures are labelled and kept apart.
• A written corrections policy with a seven-day target.

• It measures what agencies publish, not what they seize. Publishing habits change, and one office (Mumbai Customs) dominates one state.
• Most state police and many border-force seizures never reach PIB. Several agency websites could not be read from our network.
• Rules can attach a number to the wrong item or read a route as the seizure place; each row links to its source to check.
• Quantities and values are copied as published and are not audited. English releases only.
• The news layer is a sample from GDELT, fetched over several runs because of rate limits.

Newsrooms and research groups can ask for a custom extract, a feed of new events, or help reading a release. press@trinetra.in