On March 5, 2026, we are excited to announce the launch of jsonschemaprofiles, a specialized toolset designed to bridge the gap between standard JSON Schema and the strict requirements of Large Language Model (LLM) providers.

As “Structured Outputs” become a cornerstone of reliable AI applications, developers have encountered a recurring frustration: a JSON Schema that is technically valid according to the spec can still trigger API errors because it uses keywords or structures unsupported by a specific provider. jsonschemaprofiles solves this by allowing you to validate and coerce your schemas locally before they ever hit an API endpoint.

What is jsonschemaprofiles?

At its core, jsonschemaprofiles is a Go library and CLI tool that implements schema-profile validation. Instead of validating data against a schema, it validates the schema itself against a provider’s supported subset of JSON Schema.

The project ships with several built-in profiles to cover the most common use cases:

  • OpenAI (OPENAI_202602): Tailored for OpenAI’s Structured Outputs subset.
  • Gemini (GEMINI_202602, GEMINI_202503): Supporting both baseline and version-specific requirements like propertyOrdering.
  • Minimal (MINIMAL_202602): A “lowest common denominator” profile designed for developers who need their schemas to work across multiple different providers.

Beyond simple validation, the tool includes a powerful Coercion Engine. If a schema is found to be non-compliant, jsonschemaprofiles can automatically transform it into a compliant version using either a Conservative approach (minimal changes) or a Permissive approach (dropping unsupported elements).

Why it matters

For any developer building production-grade LLM integrations, predictability is everything. Relying on the LLM provider’s API to tell you that a schema is invalid is a slow and fragile feedback loop.

By integrating jsonschemaprofiles into your local development workflow or CI/CD pipeline, you can:

  • Eliminate Runtime Failures: Catch incompatible schema definitions before they cause API crashes.
  • Simplify Cross-Provider Logic: Use the Minimal profile to ensure your data structures are portable.
  • Automate Compliance: Use the coercion engine to quickly adapt complex schemas to provider restrictions without manual rewriting.

Getting Started

jsonschemaprofiles is available as both a programmable Go library and a standalone CLI.

Using the CLI: You can install the CLI tool directly via Go:

go install github.com/UnitVectorY-Labs/jsonschemaprofiles/cmd/jsonschemaprofiles@latest

Once installed, you can list available profiles with jsonschemaprofiles profiles list or validate a schema using jsonschemaprofiles validate schema <file>.

Using the Go Library: Integrate the validation and coercion logic directly into your Go applications:

go get github.com/UnitVectorY-Labs/jsonschemaprofiles

Pre-built binaries for Windows, macOS, and Linux are also available on our GitHub Releases page.

We believe that making structured AI outputs more predictable will accelerate the development of complex AI agents and applications. We invite you to try jsonschemaprofiles and help us expand the library of supported profiles.

Transparency Note: This post was AI-generated using the model unsloth/gemma-4-31B-it-GGUF:UD-Q5_K_XL.