Reliable transforms
Format, validate, convert, and compare data without generated scripts or hand-edited payloads.
Deterministic compute for agents
One HTTP API that formats, validates and converts JSON, CSV, YAML and XML, compares two documents, and handles Base64, URL, HTML entity and JWT decoding. Start with familiar developer tools, then automate the same contracts when you need to.
Illustrative estimate. Savings vary with payload size and integration; keeping large payloads outside the model context can save more than sending them as tool arguments.
One request
Every operation uses the same envelope: a versioned URL, a JSON body, an explicit result, and a stable error shape. No prompt engineering and no parsing prose back into data.
curl -X POST https://api.jsonkit.tools/v1/json/to-csv \ -H "Content-Type: application/json" \ -d '{ "input": "[{\"id\":1,\"agent\":\"Atlas\"}]", "responseMode": "plain" }'
id,agent 1,Atlas
Built for automation
Language models are excellent planners and expensive serializers. Deterministic utilities keep routine transformations fast, repeatable, and easy to inspect.
Format, validate, convert, and compare data without generated scripts or hand-edited payloads.
Every endpoint returns ok plus either result or a structured error with a machine-readable code.
Use a typed JSON envelope in applications or request the clean transformed document with no wrapper.
Requests are transformed in memory. The API design does not require storing user payloads.
The API is designed for Cloudflare's global network, with small handlers and no round trip to a database.
The version lives in every URL. Breaking changes belong in a new namespace, never a silent response change.
The full surface
Pretty-print or minify a document, catch the line that breaks it, check it against a JSON Schema, move it between formats, or compare two versions by structure instead of by whitespace. Encoding, hashing and identifier helpers cover the rest of the everyday work. Each name below opens its request and response contract in the reference.
Before your first call
Short answers to what usually decides whether an API is worth wiring up. The reference has the rest, and /openapi.json has the machine-readable version of all of it.
Twenty-eight operations over HTTP: format, minify and validate JSON, validate it against a JSON Schema, convert between JSON, CSV, YAML and XML, generate TypeScript interfaces from a payload, diff two JSON, YAML or XML documents, encode and decode Base64, URL escapes and HTML entities, decode a JWT, hash text, generate UUIDs, parse timestamps and convert colors.
No. There is no signup and no key. CORS allows every origin, so a browser, a script or an agent can call it directly. A rate limit applies per IP address to keep the service available.
POST your document to /v1/json/to-csv with { "input": "…" }. Add "responseMode": "plain" and the response body is the CSV itself, with no JSON wrapper to unpack. The same pattern works for YAML, XML and TypeScript output.
No. Every request is parsed, transformed and answered in memory. Nothing is written to disk or to a database, and there is no account to attach a payload to.
Same tested core, different delivery. The website runs it in your browser for a document you have in front of you; the API runs it over HTTP for code, CI pipelines and agents that need the same answer without a person in the loop.
Yes. /openapi.json is an OpenAPI 3.0 document generated from the same catalog that renders the reference, so every endpoint, option, default and error code in it is the one the service actually enforces.
A request body may be up to 1 MiB, a response up to 4 MiB, and parsed JSON may nest up to 100 levels. Exceeding a limit returns a structured error, never a truncated document.
JSON is only the start
/json/format/json/validate/json/diff/json/to-csv/csv/to-json/yaml/to-json/xml/to-json/json/to-typescript