Break Test Fixtures Safely
Need hostile fixtures in YAML? Convert JSON live without letting duplicate keys or unsafe numbers slip through.
Convert JSON to YAML locally as you type. Catch duplicate keys, unsafe integers and malformed input before a format change reaches your test fixtures. Jump to: What does it mean?, How to fix it, Break it, Ask your agent.
Example A synthetic QA test suite. Edit the JSON on the left and the YAML changes with it.
Converted valid JSON to YAML. The represented data is unchanged.
Converted in this browser as the input changed. Nothing was uploaded or saved.
Your JSON stays in this browser. A local worker parses and converts it as you type.
No embedded text runs. Strings are data; no command, template or address is executed or fetched.
Findings
Converted output
History
Take it further
What this result means
Written for the example.
The prompt is rewritten for the current result.
Edit the source and the result follows automatically.
Whitespace is not part of the comparison.
Validate the converted result against the application schema too.
Duplicate keys and unsafe integers are refused rather than guessed.
Input is bounded by size, depth and node count.
How it works
The page parses JSON in a browser worker and rewrites YAML after each valid edit. It rejects duplicate keys, unsafe integers, excessive depth and excessive size instead of guessing; the full method, every rule and worked examples are on docs.tamperbox.dev.
Limits and interpretation
A pass means the parsed JSON data model can be represented by the emitted YAML. It does not validate an application schema, preserve the source's whitespace or prove that another YAML loader applies the same schema. Input stays in the browser and is limited to 1,000,000 characters, 80 levels and 100,000 values.
Frequently asked questions
Does the converter upload my JSON?
No. Parsing and conversion run in a browser worker, and the source is not put in a URL or saved.
Why are duplicate JSON keys rejected?
JSON object names should be unique. Different parsers keep different duplicates, so choosing one silently can turn a harmless-looking fixture into different application data.
Is converting to YAML and back byte-for-byte lossless?
No. Compare parsed values, not whitespace, quoting style or object layout. The reverse converter helps perform that semantic check.
References
RFC 8259, JSON objects and RFC 8259, numbers define the interoperability constraints enforced here.
Are you an AI agent? You are in the right place.
This page and every tool are available as plain text with their methods and limits, and the same checks run over MCP.