Skip to article content

Revised 19 October 2026 · v1.0.0 · what changed

Code: github.com/herrrickshaw/agri-commodity-tracker

Watch the walkthrough

Watch on YouTube: youtu.be/0P80NVYwPCM

Why it is useful

• Analysts, journalists and students who need market-level (not just state-average) Indian crop prices get a daily history without paying for a data feed or registering for a portal.
• It replaces hand-downloading Agmarknet pages: one script pulls the day, one table keeps the history, and re-running a day replaces that day instead of duplicating it.
• The repo already ships CSV snapshots, so the demo needed no network at all: 463,023 rows across 21 Jul to 24 Sep 2026 loaded and queried in seconds.
• The run also showed a trap worth knowing: a naive commodity-average day-on-day comparison put 'Wood' at -39.8%, while a like-for-like market comparison showed prices essentially flat. The post shows both.

What it is for

agri-commodity-tracker collects APMC mandi prices (state, district, market, commodity, variety, minimum, maximum and modal price in rupees per quintal) and a cross-country agriculture indicator panel from open sources, and writes them to data/agri.duckdb. Committed CSV snapshots let you explore the data without running any collector.

It also has collectors for the FCI depot directory, FAOSTAT producer prices and crop production, and NASA POWER monthly rainfall and temperature for 14 Indian states.

Who it is for

• Analysts and researchers studying Indian food price behaviour
• Students learning DuckDB on a real dataset
• Developers building commodity dashboards or alerts

What you need

Get the code and an AI assistant

You do not have to be a programmer. Get the repository, give it to an AI assistant, and ask in plain language. Any of these routes works:

git clone https://github.com/herrrickshaw/agri-commodity-tracker.git
cd agri-commodity-tracker
claude      # or: gemini

Install

1. Clone and enter the repo

   git clone https://github.com/herrrickshaw/agri-commodity-tracker.git
   cd agri-commodity-tracker

2. Create a virtual environment and install dependencies

   python3 -m venv .venv && source .venv/bin/activate
   pip install -r requirements.txt pytest

Quick start: commands and what you should see

Run the test suite

python -m pytest -q

10 passed in about 5 seconds.

Collect today's mandi prices (live; needs network, slow with the demo key)

python3 collectors/mandi_prices.py

Rows are written to data/agri.duckdb, replacing any earlier pull for the same date. Not run in this demo, see limits.

Explore the committed snapshot without any network, with DuckDB

python3 -c "import duckdb; print(duckdb.sql(\"select pulled_on, count(*) n, count(distinct commodity) commodities from read_csv_auto('data/mandi_prices_snapshot.csv') group by 1 order by 1 desc limit 3\"))"

A table of the latest pull dates with row and commodity counts.

Sample outputs

Real output from a run on the date shown in the sources line; trimmed to fit.

Coverage of the latest pulls in the committed snapshot

select pulled_on, count(*) n, count(distinct commodity) com, count(distinct state) st from m group by 1 order by 1 desc limit 4
pulled_on    n      com  st
2026-09-24   13375  227  28
2026-09-23   11583  224  24
2026-09-21   11258  221  24
2026-09-20    7987  151  21

Whole file: 463,023 rows, pulled 2026-07-21 to 2026-09-24

Days differ a lot in size because the source feed fills through the day.

Naive day-on-day movers (average of all markets, 24 vs 23 Sep 2026, commodities with 30+ rows)

average modal_price per commodity per pull date, then percent change
commodity       prev   latest  pct
Wood            1005    605    -39.8
Pineapple       4978   5779    +16.1
Tender Coconut  4209   4617     +9.7
Rice            4410   3998     -9.4
Water Melon     1840   1969     +7.0

Prices in rupees per quintal. These big moves are mostly a change in which markets reported that day.

Like-for-like check: same state, market, commodity and variety on both days (median percent change)

join the two days on state+market+commodity+variety
commodity   same_markets  prev   latest  median_pct
Soyabean    73            6040   5894    -0.4
Bajra       30            2400   2435    +0.2
Cotton      35            7967   7978     0.0
Garlic      105           17213  17177    0.0

The same data shows essentially flat prices once the market mix is held fixed. Always compare like with like.

Highest Tomato modal prices on 24 Sep 2026

select market, state, modal_price ... where commodity='Tomato' order by modal_price desc limit 3
Manubazar (Tripura)      6200
Damnagar APMC (Gujarat)  5000
Bolpur APMC (West Bengal) 4960

Rupees per quintal.

Ways to use it

Sample prompts for your AI assistant

Copy, paste and edit the parts in [brackets].

• Explain this repository in plain English: which data sources does it collect and what tables does it create?
• Using data/mandi_prices_snapshot.csv, show the price trend of [commodity] in [state] by week, and flag days with unusually few rows.
• Compare day-on-day price change for [commodity] using only markets that reported on both days, and explain why that is safer than averaging.
• Read DATA_ACCESS.md and tell me what I need to do to run the daily collector with a personal data.gov.in key.
• Join the World Bank panel with mandi prices to show how India's agricultural value added share compares with [country], and note the limits of that comparison.

Limits and honest caveats

Sources: the agri-commodity-tracker README, DATA_ACCESS.md, tests and data/mandi_prices_snapshot.csv; pytest and DuckDB queries run on 19 October 2026 on a fresh clone.

About this article: Researched, written and edited by Umashankar Triplicane Dwarakanathan, with AI research assistance; every figure is meant to trace to the primary source cited. See the Editorial Policy for how sourcing, AI use and corrections work.