← SchemaMedic

Clean CSV, JSON and JSONL in one API call.

For developers and autonomous agents building importers, CRM pipelines, crawlers and data workflows. No account or recurring subscription is needed: full requests use x402 micropayments on Base. All transformations are deterministic and bounded.

1. Convert CSV to clean JSON

curl -X POST https://schema-medic-eclipselio.onrender.com/v1/preview \
 -H 'Content-Type: application/json' \
 -d '{"format":"csv","output":"json","input":"name,city\nAda,Athens\nAda,Athens\nBob,Patras"}'

The preview is free and shows up to three normalized records plus row counts, duplicate counts, column names and output SHA-256. Try it from the interactive demo before paying.

2. Remove duplicate rows, clean whitespace, normalize headers

Switch format among csv, tsv, json and jsonl; choose output from csv, json and jsonl. Optional Boolean settings are dedupe, trim and dropEmpty, enabled by default. Duplicate rows are compared after normalization, without changing substantive values.

3. Request the full paid result in USDC

POST https://schema-medic-eclipselio.onrender.com/v1/normalize
Content-Type: application/json

{"format":"csv","output":"json","input":"name,city\nAda,Athens\nAda,Athens\nBob,Patras"}

An unpaid call returns HTTP 402, not a paid result. An x402-capable buyer client can authorize $0.02 USDC on Base mainnet and retry the same request with its payment signature. After facilitator verification/settlement, the complete response is returned automatically. This does not require the seller to spend gas or operate a custodial wallet.

4. Audit schema quality and detect drift

For imports and ETL/agent pipelines where a conversion alone is insufficient, call POST /v1/profile. It costs $0.05 USDC on Base, paid via the same x402 V2 flow.

POST /v1/profile
Content-Type: application/json

{
 "format":"csv",
 "input":"id,amount\n1,20\n1,20\n2,",
 "baseline":{"fields":[
   {"name":"id","type":"number","required":true},
   {"name":"amount","type":"number","required":true}
 ]}
}

The report includes per-column missingness, observed types, distinct value counts, numeric minimum/maximum/mean, exact duplicate counts, row-shape warnings and comparison with the optional baseline. Type inference is heuristic: do not treat it as semantic validation, an AI analysis or a guarantee of data quality.

Who this fits

Agents repairing inconsistent exports, data processors converting CSV to JSONL, ETL workflows that need exact deduplication, and services that need reproducible data transformation with a content digest.

What this does not do

It cannot recover lost columns, guess malformed JSON meanings, promise AI quality or guarantee buyer purchases. Do not send passwords, private keys or regulated data. Free Render cold starts may apply.

Full API specification · Agent integration skill · x402 discovery