Better tools for the data between systems.

Why we built it

JSON is simple enough to read and subtle enough to fail in expensive ways. A misplaced comma stops parsing; a large integer can lose precision; duplicate keys can be interpreted differently; a useful debugging payload can expose a live credential. JSON Anvil focuses on those real situations rather than presenting a blank textarea with no explanation.

Each tool is designed to produce a reviewable result. Inputs remain visible, errors are not hidden, and guides explain where automated output needs human judgment.

Local-first by design

The JSON entered into our formatter, repair, diff, JSONPath, schema, type-generation and redaction tools is processed inside the browser. JSON Anvil does not upload tool input to an application server or create user accounts.

Running locally is useful for privacy, speed and resilience, but it is not a substitute for careful handling. A browser extension, copied output or third-party software on a device can still create risk. Avoid pasting production secrets when a representative fixture will do.

Editorial approach

Guides are written around concrete, reproducible behavior. We separate syntax validity from schema and business validity, call out implementation differences, and avoid claiming that one example can define a complete contract. Pages are reviewed for clarity and updated when standards or common runtime behavior changes.

Examples use reserved domains and documentation address ranges where possible. Automated assistance may support drafting and editing, but JSON Anvil remains responsible for the published result.

What comes next

The initial workshop covers formatting, repair, structural comparison, JSONPath, schema inference, TypeScript generation and safer redaction. Future additions will prioritize developer problems that can be solved deterministically in the browser, including large-document inspection and schema validation.