Conservative shape inference
Maps observed JSON values to schema types without guessing dates, emails, UUIDs or constraints that one sample cannot establish.
Day 033 / APIs / JSON Schema
JSON Schema Starter
/ Overview
A dependency-free Python CLI that turns one bounded local JSON object into a deterministic Draft 2020-12 starter schema for human review.
An example API payload is useful evidence, but manually translating nested objects and arrays into a first JSON Schema is repetitive. A single sample also cannot prove every optional field, string format or business rule.
/ Provided servicesClear thinking.
Clear thinking.
Practical delivery.
Reviewable object contracts
Sorts properties, marks observed fields as required and closes objects by default, with an explicit option to allow additional properties.
Honest evidence boundary
Treats empty and mixed arrays cautiously and documents that the generated result is a starting point, not a proven production contract.
/ Solution
JSON Schema Starter reads one bounded local JSON object and recursively creates a deterministic Draft 2020-12 starting schema. It distinguishes booleans from integers, handles nested objects and array shapes, rejects unlisted properties by default and clearly documents where human review is still required.
/ Impact
Draft 2020-12 starter output
Recursive object and array inference
Deterministic property ordering
Bounded local-only input
No speculative string formats
Twelve standard-library tests included
Python standard library
Argparse, pathlib, JSON, math and deterministic serialization provide the complete dependency-free workflow.
JSON Schema
Produces a Draft 2020-12 declaration with recursively inferred object, array and scalar shapes.
Unittest
Twelve tests cover scalar distinctions, nested objects, arrays, deterministic ordering, depth limits, invalid JSON and CLI output.