Skip to content
OnchainQueries

OQ — S10 · For analysts & researchers

You already know how to write the query. What you do not want is to write an indexer first. 37 Solana programs are already decoded into 1,074 typed tables with the program's own field names.

Programs
37
Tables
1,074
Setup time
0
Credits / mo
50,000
OQ — S10·1What you get

No node to run, no schema to design, no backfill to babysit.

Typed columns

from the IDL

sol_amount, is_buy, mint — the program's real fields, not a generic decoded-log blob you have to reverse engineer.

OnchainQL you know

date_trunc, count_if

The common analytical surface ports directly, and unsupported forms fail by name rather than silently differing.

Typed parameters

{{start_time}}

Parameterise a query once and re-run it across ranges without editing SQL text.

Reproducible results

catalog hash

Every result records the tables read and the slot it was planned at, so a number can be re-derived later.

Export to your tools

Parquet · Arrow

Pull results into pandas, Polars or DuckDB and finish the analysis where you are comfortable.

Publish the finding

dashboards

Turn the query into a chart others can read, with the query still attached underneath.

OQ — S10·3Example

Hourly buy and sell flow for a launchpad, parameterised on the window.

flow.sql
SELECT  date_trunc('hour', ts) AS hour,  count_if(is_buy) AS buys,  count_if(NOT is_buy) AS sells,  sum(sol_amount) AS sol_volumeFROM pump_fun_solana.pump_fun_evt_trade_eventWHERE ts >= {{start_time}}GROUP BY hourORDER BY hour;
OQ — S10·4Questions

Do I need to know the engine underneath?

No. You write OnchainQL and OnchainQueries translates the forms it supports. Some edge-case behaviour follows the execution layer rather than the spec you are used to, which the SQL contract documents explicitly.

Can I port my existing analytical queries?

The core analytical surface ports directly. Do it incrementally — verify the table coverage first, then add functions, joins and windows one at a time, because anything unsupported fails with the feature named.

What if the table I need is not there?

Check the dataset catalog first — there are more tables than most people expect, including instruction tables. If a program is genuinely missing, it can usually be decoded from its IDL.

Fourteen days on Pro. 50,000 credits a month — five terabytes scanned.

Start queryingRead the docs