网页
综合
更新时间 2026-11-14 09:41:18
The JSON Schema Registry | Sourcemeta One
自动蜘蛛
www.sourcemeta.com
正文约 6000 字
Enterprise-grade | High-performance | Self-hosted
Build a canonical library of schemas for every OpenAPI spec in your organisation to reference. Without one, you're governing the 20% of your specs that's boilerplate and ignoring the 80% that matters.
Designed by a member of the JSON Schema Technical Steering Committee
Our technology is trusted by national laboratories, energy infrastructure providers, and regulated enterprises worldwide
You adopted OpenAPI to standardise your APIs, but your governance problems didn't go away. Your interfaces are consistent, yet five teams still define "Customer" differently.
Our guide walks you through a proven methodology for schema governance, adopted by mature engineering organisations.
Group schemas by team, department, domain, or vendor, as deep as your organisation requires. Every layer gets its own health score and contact details, so quality and ownership are visible and actionable across your entire structure.
Every schema gets a live quality score from our schema linter, the most advanced one in the market. Things like unused definitions, missing descriptions, anti-patterns, and much more are surfaced automatically and aggregated up through your organisational structure.
Your schemas live in Git, and so does your registry configuration. Every schema change and registry deployment goes through your existing review workflow. Database-backed registries force you to rebuild approval processes, access controls, and audit trails from scratch inside the registry itself. With Sourcemeta One, you get all of that out of the box, using the Git tools and processes your team already knows.
Sourcemeta One ships with a fully featured web explorer for navigating and understanding your schemas. But it is entirely optional. Run the registry headless for API-only deployments, or build your own frontend on top of our complete HTTP API. Everything you see in our UI is powered by the same API you have full access to.
Sourcemeta One exposes your entire registry as a Model Context Protocol server out of the box. Every schema in the catalog becomes an MCP resource, and every HTTP API action becomes an MCP tool. AI agents can discover, retrieve, and validate against your organisation's schemas with no bespoke glue code.
Annotate your schemas with our JSON-LD extension vocabulary and the registry will promote any valid instance to JSON-LD, the JSON serialisation of RDF. Validation and semantics run in a single pass, with full support for every conditional and dynamic JSON Schema operator, something no other tool does. The same schema that validates your data declares what that data means, with no separate context to keep in sync.
A registry is only valuable if your schemas actually reach the projects that need them. Most teams end up writing custom scripts to pull schemas into their applications, with no integrity checks, no standard workflow, and no way to handle schemas that reference other schemas. Our CLI gives you a full client-side schema management experience inspired by package managers like npm and pip, complete with dependency resolution, automatic de-referencing, checksum verification, and repeatable installs.
JSON Schema keeps publishing dialects with breaking changes, and OpenAPI adds its own on top, so your schemas and the tools that consume them are rarely on the same one. Store each schema once and let every consumer ask for the dialect it understands on the fetch URL, rewritten and bundled on the fly. We mapped how fragmented this really is, and why no other upgrader gets it right.
Written in C++ and powered by Blaze, our high-performance JSON Schema validator. Each instance starts in under 2 milliseconds and requires zero external dependencies. No database, no message queue, no JVM. Just a single stateless Docker container that scales horizontally without limits.
The table below covers the most commonly evaluated alternatives.
One path not shown is building a registry in-house. While you may reach a V1 faster, it introduces ongoing maintenance burden, carries high long-term development costs, and rarely matches the governance capabilities of a solution designed by JSON Schema experts.
Comes with world-class support, directly from a member of the JSON Schema Technical Steering Committee.
We set up a Slack Connect channel, or whichever tool you prefer, so your engineers reach us directly and get fast answers day to day, as if we sat on your own team. No ticket queue, no first-line agent reading from a script.
We shape your schema strategy, teach JSON Schema best practices, and review your pull requests whenever you want a second set of expert eyes.
Want it off your plate entirely? We take full ownership of schema authoring projects and can run your whole schema governance programme for you, end to end.
A perfect fit for teams where security, compliance, and operational independence are non-negotiable.
The Sourcemeta One Enterprise edition, while not open source, is fully available on GitHub for security audits, compliance reviews, and internal assessments. No black-box concerns, no blind trust required. If Sourcemeta ever ceases to exist, you retain full rights to continue running and maintaining your own instance.
Every release ships with signed container images via Cosign and Sigstore, SLSA Build Level 3 provenance, and a signed SPDX Software Bill of Materials. Cryptographic operations are built against the OpenSSL FIPS provider for industries where compliance is non-negotiable.
Your schemas live as JSON Schema, an open and widely adopted standard, not a proprietary language like Looker's LookML or Palantir's Foundry Ontology that only runs inside one vendor's platform. Every definition stays a portable, vendor-neutral file you can read, validate, and run anywhere, even if you ever leave us.
Built in C++, proven by research to be at least 10x faster than alternatives. Each instance starts in under 2 milliseconds. Stateless by design and fully independen
本页为82搜索生成的网页快照,内容来自站内索引;演示数据的原文链接仅作示例。