Permits, wells, leases, operators, production and inspections — cleaned, joined across every file, and traceable to the exact state record it came from. One API for your analysts and your AI agents.
That knot in your stomach isn't paranoia. It's what working with raw RRC data feels like.
An operator's wells, permits, inspections and violations. A lease's full production history since 1993. No spreadsheets, no joins, no weekend.
Each response names the snapshot and the source releases behind it. Every row links back to the original RRC file by its SHA-256 hash.
Self-describing catalog, consistent IDs and read-only SQL, so an AI agent gets the right answer instead of a confident wrong one.
Original files are pulled from the RRC on its own schedule — nightly, weekly or monthly — and stored byte-for-byte, never overwritten.
Each file becomes columnar data. A proof rebuilds the source from the output, so nothing is dropped or invented.
Operators, wells, leases, permits, fields and districts get one consistent ID across every file. The original value stays right beside it.
Releases are frozen into a hash-identified snapshot and kept. Every answer names its snapshot, so any result can be reproduced and audited.
Live today: drilling permits (master, daily and horizontal), pending permits, P-5 operators, production (PDQ), pending production reports, wellbore query, inspections and violations. Gas and oil proration ledgers, field rules and full wellbore history are captured and converted, and are being added to the API.
Every answer carries its provenance, so your agent can cite it and your auditor can check it. Decimals come back as exact strings — never rounded through a float.
And it's fast. Measured server time: a permit lookup in 10 ms, a full year of statewide production in 3 ms, an operator's entire production history in 6 ms.
Operators, wells, leases and permits, each assembled across every dataset.
Monthly oil, gas, condensate and casinghead, statewide or by district, operator or field, since 1993.
Query any table. Sandboxed, time-limited, row-capped.
Plug straight into Claude, ChatGPT or your own agent.
# Monthly production for one oil lease curl https://api.texasrrcapi.com/v1/leases/O-01-00001 \ -H "X-API-Key: $KEY" { "data": { "production": { "rows": [ { "CYCLE_YEAR_MONTH": "199301", "LEASE_OIL_PROD_VOL": "…", "lease_id": "O-01-00001", "operator_id": "…" }, … ] }, "wellbores": { … }, "inspections": { … } }, "meta": { "snapshot_id": "cc88b9bc…", "sources": { "pdq": { "source_release_id": "102afef5…" } } } }
| Download it yourself | Enterprise data platforms | texasrrcapi | |
|---|---|---|---|
| Cost | Free, plus your weekends | Enterprise contracts | From $0, Pro at $299/mo |
| IDs joined across files | You build it | Varies by vendor | Yes, raw value kept beside it |
| Trace a value to the source file | If you saved it | Varies by vendor | Every row, by hash |
| Frozen, citable snapshots | State files are overwritten | Varies by vendor | Every response names one |
| Built for AI agents | No | Varies by vendor | Designed for it |
Try every dataset.
For analysts and single agents.
For teams and products.
For firms that embed or redistribute.
No. texasrrcapi is independent and not affiliated with the Railroad Commission of Texas. It is built entirely from the Commission's public data files, and every record links back to the file it came from.
Each source is refreshed on the Commission's own schedule: pending permits daily, inspections and violations weekly, production and most master files monthly. Every response tells you exactly which release you're looking at.
Never. Raw values are kept exactly as the state published them. Cleaned IDs, typed numbers and dates are added beside them, never in place of them. Decimals are returned as exact strings.
Ten datasets are live today. More structured files are being added, and scanned documents such as permit PDFs will follow. Placeholder values in the state's data (like "Unknown" operators) are kept as-is and flagged rather than guessed.
Bulk Parquet downloads are coming on the Business plan. Until then, read-only SQL lets you pull exactly the slice you need.