Have Your AI Analyze the Cricket Data on Tigzig. 24+ Years of Ball by Ball, ODI and T20, Across 6000+ Matches.
Published: August 22, 2026
Did you know you can have your AI analyze the cricket data on Tigzig - 24+ years of ball-by-ball, ODI and T20, covering 6000+ matches.
Paste this into whichever AI you use:
go to api.tigzig.com and get me some high level diagnostics on the cricket
database there and the kind of analysis we can do
That address is a machine-readable catalog, so the agent reads it, works out which service holds the cricket data, opens that service's OpenAPI spec, finds the tables and their columns, and writes the queries itself.
Free and open, no signup, refreshed twice a day.
Docs Hub: tigzig.com/apis/database
Credit: Data sourced from Cricsheet, ODC-BY 1.0.
Which AI tools this works with
Works with any AI tool with API access, which includes Claude AI chat, Claude Cowork, ChatGPT Work as well as all AI Coders like Claude Code, Antigravity, Codex. The regular ChatGPT can't access it but if you have a ChatGPT Go or Pro subscription it comes with ChatGPT Work. Some useful guides here:
- Claude Cowork: Working with Claude Cowork and Claude Code
- ChatGPT Work: ChatGPT Work comes with your Go or Pro subscription
- Google Antigravity: Antigravity, a full AI coding agent
What people ask, and what the schema gives you
Since it has the schema, the questions people tend to ask are strike rate against pace versus spin, run rates through the death overs, head to head between two teams across a decade, or how a particular venue behaves batting first.
There are three tables on each engine, odi_ball_by_ball, t20_ball_by_ball and match_info, all joined on match_id. match_info carries venue, dates, teams, toss and result, so you can slice by any of those without pulling in a second source.
Because it is ball by ball rather than pre-aggregated, anything you can express in SQL over deliveries is open to you, including the very specific cuts you want to pull up.
The SQL is quite permissive, so multi-table joins and multiple CTEs will run, and your agent can draft all of that from a plain description of what you are after, with the API docs setting out which constructs are allowed and which get refused.
On validations
As for validations, the data ingestion pipeline has validation checks built in at multiple levels, and a new set now runs against the final tables on both engines, compared row for row, with every run recorded pass or fail and kept with its history. It covers referential integrity in both directions, null density, freshness, and cricket validity, things like no T20 innings past over 19 and no delivery scoring below zero or above seven off the bat.
The first run found something. The two engines disagreed by one row, a Test match present in one and missing from the other. That is fixed.
The page that publishes all of this is still being designed and will go up soon. Anomalies that remain will be published as found rather than cleaned up. For example some delivery slots hold two rows, which is mostly a wide or no-ball being re-bowled at the same over and ball number, and the rest will be shown as unexplained. That remainder is fewer than 200 slots out of more than 2 million deliveries.