Confluence and APIMatic are the two providers the Documentation tag page ranks first, at 67.4 and 60.6 on the Kin Score, exemplar against strong, 6.8 points apart. The tag holds 84 providers and 133 APIs with a mean of 32.8, so both sit well above their category. The Kin Score measures the public API surface, not the product and not the company, and this pairing needs that clause more than most, because the two are not the same kind of documentation. Confluence is where a team writes its docs. APIMatic is what turns an API specification into docs, SDKs, and now MCP servers. A buyer is rarely choosing between them. A reader of this series should still see how they score, because APIMatic wins five of the six facets and loses the composite on the one it did not publish.
The business. Confluence is Atlassian’s wiki and knowledge management product, and the record carries its pricing in full: free up to 10 users, Standard at $6.05 per user per month, Premium at $11.55, Enterprise typically $23 to $25, with a REST API at api.atlassian.com, OAuth 2.0 three-legged auth and API tokens, webhooks for content changes, CQL for search, and the Connect and Forge frameworks for marketplace apps. APIMatic is a developer experience platform for APIs: SDK generation in seven-plus languages, documentation portals, specification validation and linting, format transformation across fifteen-plus specification formats, an OpenAPI linter as a GitHub app, and MCP server generation. Both are hosted and self-serve. Confluence is freemium; APIMatic is paid with a 14-day trial. Confluence documents 2 rate limits and 9 OAuth scopes; APIMatic documents 7 rate limits and 1 scope. Confluence’s security record lists SOC 2 Type II, SOC 3, ISO 27001, 27017, 27018 and 27701, PCI DSS, FedRAMP, HIPAA, CSA STAR, and GDPR, plus a published security.txt. APIMatic’s lists TLS 1.3, HSTS, and DMARC.
The APIs. Confluence publishes 117 API pages to APIMatic’s 5. Confluence’s are carved by the catalog from Atlassian’s own swagger.v3.json, one page per tag, so the Confluence Content API, the Confluence Search API, and the Confluence Space API are Atlassian’s contract read three ways, alongside a Confluence GraphQL API and a separate Data Center REST API for on-premise. The record holds 55 contracts, none derived by us, but only 10.9% of them callable as published, because the servers block carries a tenant placeholder rather than a host. The 13 workflows and 43 examples on the record carry no first-party marker, so the post does not count them. APIMatic’s side is the disclosure. All 11 of its contracts were generated by API Evangelist from the structured reference data its own documentation portal loads, because the portal’s spec export is disabled and no OpenAPI for the Platform API was found on APIMatic’s hosts or in its GitHub organization. What APIMatic does publish first-party is an MCP server, two agent skills, and its conformance documentation. Confluence’s MCP server is first-party too.
The experience, facet by facet. APIMatic takes five: discoverability 81.5 to 74.1, contract governance 33.3 to 18.2, operational transparency 57.9 to 44.7, access clarity 77.6 to 63.2, and developer ergonomics by a hair, 95.2 to 94.6. Confluence takes one, contract quality, 71.1 to 28.4, a gap of 42.7 points. Contract quality carries 0.25 of the composite, the heaviest weight in the rubric, and APIMatic’s five facets together carry 0.75 but none of them by more than 15.1 points. On the six-facet base Confluence leads 64.7 to 61.3, and the conditional layers widen it slightly, plus 2.7 for Confluence on an upsert score of 33.3 against minus 0.7 for APIMatic on 0.0. Neither triggers the regulatory facet, so the composites are computed over the same facet set, which makes this the cleanest comparison the series has run.
The asterisk is on the one facet, not the ranking. APIMatic’s 28.4 for contract quality is the quality of contracts we generated from its documentation, not of a contract it wrote, so the number is a measurement of our derivation. What it measures about APIMatic is the absence: a company whose product is turning specifications into developer experience does not publish a specification for its own Platform API, and has export switched off on the portal that would produce one. The catalog gap block says the same thing in a different way. APIMatic has earned 93.0 of a possible 115.0 points across the artifact classes, more than Confluence’s 49.5, and its first-party share is 24.0 to Confluence’s 0.0. On everything a provider can publish about itself except the contract, APIMatic is the more complete record. The contract is the piece a spec-tooling company should find easiest.
Agent readiness is close, Confluence 54.4 to APIMatic 48.8, both agent-ready, neither agent-native because idempotency is unlit on both. The identity layer is where this pairing is unusual: both light delegated identity, both have protected resource metadata verified, and both support dynamic client registration, in a category where 11.7% do. Both document an MCP server, both a rate limit signal, both have error semantics verified. Confluence adds partial OpenAPI examples, documented reversibility, and a described event surface. APIMatic adds first-party agent skills, which Confluence has only as a derived artifact. Against a documentation cohort where 49.4% have agentic access and 74.0% a rate limit signal, both are near the top of the category rather than exceptions to it.
The verdict is the one the shapes give. If you are writing to an API that stores documents, Confluence’s contract is the better one, and it is Atlassian’s own, with the most complete governance and security record in the pairing. If you are choosing a platform to make your own API’s documentation and agent surface, APIMatic’s record shows a company that has done most of that work on itself: first-party MCP, first-party skills, full identity layer, and better transparency, governance, and access clarity than the market leader in wikis. What would move APIMatic’s number is one file. Publish the Platform API’s OpenAPI, or turn export back on in the portal, and the heaviest facet in the rubric stops being scored on our derivation. What would move Confluence’s is the four facets it loses by double digits, starting with contract governance at 18.2, where a changelog and a deprecation policy on the contract it already publishes would do most of the work.