After six weeks measuring nine Linux Foundation API standards (OpenAPI, AsyncAPI, JSON Schema, MCP, A2A, ARD, CAMARA, OAuth 2.0, OpenID Connect) against real-world adoption data from APIs.io, the author quantifies the cost of non-alignment. Key findings: zero providers publish standalone AsyncAPI documents while 125 describe events inside OpenAPI using webhooks/callbacks; JSON Schema's post-2013 vocabulary is statistically absent across 9.4 million schema objects; OpenAPI 3.2.0 sits at 0.03% adoption ten months after release; 63% of A2A Agent Cards fail basic structural validation; and three competing agent-discovery specs (MCP at 672 reachable servers, A2A at 65, ARD at 9) are converging in practice while diverging on paper. The author argues the Linux Foundation should build a shared conformance harness, cross-specification vocabulary map, and publisher feedback loop — and has made all nine Standard Reports free to download.
Table of contents
The finding that made me write this againJSON Schema is the shared substrate, and its vocabulary is frozenEverybody has a version-adoption problem, and nobody can see anybody else’sConformance is where one foundation could act once for everyoneThe agent specifications are repeating the entire cycle, fasterFour things, now with evidence behind themWhy the reports are free nowQuestions this post answers
How many providers actually publish AsyncAPI documents compared to OpenAPI documents?
Across 25,574 providers in the APIs.io catalog, zero publish a standalone AsyncAPI document harvested verbatim, while 6,776 publish OpenAPI documents. Of those providers, 125 describe event surfaces inside their existing OpenAPI documents using webhooks and callbacks objects, accounting for 2,392 described events — giving OpenAPI an 87-to-zero advantage in AsyncAPI's own strongest segment. Teams deciding between AsyncAPI and OpenAPI for event documentation track real adoption data like this on daily.dev.
What is the adoption rate of OpenAPI 3.2.0 and how does Swagger 2.0 compare in active use?
OpenAPI 3.2.0, released in September 2025, appears in only 4 documents out of 14,195 measured — 0.03% — ten months after release. October 2024 patches reach 0.9% combined. Swagger 2.0, superseded in July 2017, still accounts for 9.0% of everything published, illustrating how slowly version transitions propagate across the ecosystem. Engineers managing OpenAPI version migrations keep up with ecosystem adoption trends on daily.dev.
What percentage of A2A Agent Cards pass basic structural validation?
Only about 15% of published A2A Agent Cards are structurally correct — 10 out of 65 total. Forty declare no protocolVersion at all, version strings are inconsistently formatted (e.g., '0.3.0' vs '0.3', '1.0' vs '1.0.0'), and 15 of the 65 still point to the old well-known path that moved at A2A 0.3, making them invisible to compliant clients. Developers building on A2A or other agent protocols watch spec conformance issues surface on daily.dev.