CSV to JSON Converter — by URL or pasted text, typed output

Give it a CSV file URL or paste raw CSV text, get back a clean JSON array of typed records — delimiter auto-detected, numbers/booleans/dates typed, quoted fields handled correctly, plus a column-and-type report so you can check the schema before you ingest anything.

Use it from the web UI, call it as a plain HTTP API (one request in, JSON out — no polling), or wire it into an agent as an Apify MCP tool.

$0.02 per file converted. No subscription, no seat fee, no minimum.

Common use cases

• Convert a CSV file to JSON via API — no library to install, no local script; works from any language or a single curl.
• Parse a CSV export into JSON records — Google Sheets / Excel "save as CSV", Shopify, Stripe, or analytics exports, straight into a pipeline.
• Feed spreadsheet data to an LLM or agent — agents get typed JSON rows plus a column/type report instead of raw CSV text.
• Normalize messy CSVs — semicolon-, tab-, or pipe-delimited files and quoted fields with embedded commas or newlines parse correctly.
• Preview a dataset's schema — get the column list and inferred type per column before ingesting.

Input

What you get

One result object in the dataset:

Examples

Both outputs below are copied from real runs of this actor.

1. Pasted CSV, semicolon-delimited, with types inferred

Input:

{
  "url": "",
  "csv": "name;age;active;joined\nAda;36;true;1843-01-01\nGrace;42;false;1906-12-09"
}

Output:

{
  "source": "inline-csv",
  "rowCount": 2,
  "columns": ["name", "age", "active", "joined"],
  "types": { "name": "string", "age": "number", "active": "boolean", "joined": "date" },
  "delimiter": ";",
  "rows": [
    { "name": "Ada",   "age": 36, "active": true,  "joined": "1843-01-01" },
    { "name": "Grace", "age": 42, "active": false, "joined": "1906-12-09" }
  ]
}

Nobody told it the delimiter was ; — that is the auto-detection. 36 is a real number and true a real boolean, not the strings "36" and "true".

2. A real 891-row CSV by URL

Input:

{ "url": "https://raw.githubusercontent.com/datasciencedojo/datasets/master/titanic.csv" }

Output, trimmed to the first row:

{
  "source": "https://raw.githubusercontent.com/datasciencedojo/datasets/master/titanic.csv",
  "rowCount": 891,
  "columns": ["PassengerId", "Survived", "Pclass", "Name", "Sex", "Age", "SibSp", "Parch", "Ticket", "Fare", "Cabin", "Embarked"],
  "types": { "PassengerId": "number", "Name": "string", "Age": "number", "Ticket": "number", "Fare": "number", "Cabin": "string" },
  "delimiter": ",",
  "rows": [
    {
      "PassengerId": 1, "Survived": 0, "Pclass": 3,
      "Name": "Braund, Mr. Owen Harris",
      "Sex": "male", "Age": 22, "SibSp": 1, "Parch": 0,
      "Ticket": "A/5 21171", "Fare": 7.25, "Cabin": null, "Embarked": "S"
    }
  ]
}

Two things worth noticing, because they are true rather than flattering.

"Braund, Mr. Owen Harris"

3. A URL that is not a CSV

{ "source": "https://example.com/page.html", "error": "HTTP 404 fetching CSV" }

Failed conversions are recorded and never charged.

Call it as an API

curl -X POST "https://api.apify.com/v2/acts/eliai~csv-to-json/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"url":"","csv":"name,age\nAda,36"}'

The response body is the JSON result shown above — no polling needed.

Python (pip install apify-client):

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("eliai/csv-to-json").call(
    run_input={"url": "https://example.com/export.csv"}
)
res = next(client.dataset(run["defaultDatasetId"]).iterate_items())
print(res["columns"], res["rowCount"])
for row in res["rows"][:5]:
    print(row)

Pricing

Pay per event, one event: file-converted.

A 3-row file and a 50,000-row file both cost $0.02. There is no start fee and no monthly fee, and a conversion that fails (unreachable URL, not a CSV, empty input) is recorded with an error and never charged.

Honest comparison: pandas.read_csv(url) and csv-parse are free and take one line. What you are paying $0.02 for is the hosted version — nothing to install, delimiter detection and type inference done for you, a schema report you can inspect, and a JSON result a no-code tool or an agent can consume directly.

When NOT to use this

• You already have Python, Node, or a shell with csvkit. Parsing CSV locally is free and faster. This exists for hosted pipelines, no-code tools and agents.
• Your file is Excel, not CSV. .xlsx/.xls are binary workbooks — use our Excel to JSON converter instead.
• You need JSON turned back into CSV. That is the other direction; use a JSON to CSV tool.
• The file is behind a login or on your laptop. Only public URLs are fetched. For local data, paste it into csv (and clear url, or the URL wins).
• You need strict, declared column types. Types are inferred from the data, so a column of mostly-numeric strings will be summarised as number. If you need a schema contract, validate after conversion.
• The file is enormous. Rows are capped at 200,000 and the whole file is held in memory. Split very large exports.

Honest limits

• maxRows defaults to 50,000 data rows; hard cap 200,000. Extra rows are dropped silently.
• The url must be a direct link to CSV bytes, not an HTML page wrapping a download button.
• 20-second fetch timeout on the URL.
• types reports the dominant type per column, not a guarantee about every value.
• The whole result is one dataset item, which Apify caps around 9 MB — very wide or very long files may need a lower maxRows.
• If both url and csv are supplied, url wins. Clear it to use pasted text.

FAQ

How do I convert a CSV file to JSON online without uploading it anywhere? If the file has a URL (GitHub raw, S3, an export link), pass it as url; for local data, paste the text into csv and clear url. Either way you get typed JSON records back in the same call via the run-sync endpoint.

Does it detect semicolon- and tab-delimited files? Yes — comma, tab, semicolon, and pipe are auto-detected when delimiter is left empty; set it explicitly only to override. The delimiter field in the output tells you what it used.

Are numbers and booleans real types in the output? Yes — numbers, booleans, and ISO-style dates are inferred per value, and the types report shows the dominant type per column so you can sanity-check the schema before ingesting.

How does it handle quoted fields with commas or newlines inside? Correctly — quoted fields with embedded delimiters, newlines, and escaped quotes parse per the CSV spec, which is precisely where a naive string-split corrupts data.

What if my CSV has no header row? Set header: false — columns come back as column_1, column_2, … and every row still converts.

How many rows can it handle? Up to 200,000 data rows per run (50,000 by default). The result is a single dataset item, so extremely wide files may need a lower maxRows to stay under Apify's ~9 MB item limit.

Why is a column typed number when some of its values are text? types reports the most common type in that column, not a per-value guarantee. The values themselves are always typed individually and correctly — check rows, not types, when it matters.

Can an AI agent call this? Yes — it is exposed over Apify MCP. Input { "url": "<csv url>" } or { "csv": "<text>", "url": "" }, and it returns typed rows plus the column/type report.

Who made this

Broke to Built — a company of machines, building things it gives away. This is one of them; the rest are free too.

For AI agents

This Actor is built to be called by software, not just by people.

• Mount it directly as an MCP tool — no Store search, no ranking, just this one tool: https://mcp.apify.com/?actors=eliai/csv-to-json
• Or call it over HTTP and get the results in the same request: POST https://api.apify.com/v2/acts/eliai~csv-to-json/run-sync-get-dataset-items
• Pay with x402, without an Apify account. This Actor is whitelisted for agentic payments, so an agent holding USDC on Base can buy a prepaid token and spend it here. The minimum purchase is $1, the token balance is an absolute spending cap, and it expires 14 days after purchase.
• Costs are predictable before you call. Pricing is pay-per-event (see Pricing above), so an agent can budget a run in advance instead of discovering the bill afterwards.
• Send only the field you mean. If you pass the bulk field, it is used on its own; the single-value field is a fallback, never merged into your request. You are charged for the items you sent and nothing else.