Docs platforms

Docs platforms and AI agent readiness

Grouped by the tool serving the docs.

The AI Agent Readiness Test records which tool serves each product’s documentation, then scores the docs themselves. The platform is never scored. Grouping the results shows which platforms tend to ship the agent-facing surface their users score on. Every platform seen in a published report appears here, and the product count next to each one says how much weight the numbers carry.

21 platforms across 97 reports

Every documentation platform seen in a published report.
Platform Products Average score Range
Mintlify 32 90.2% 73.8% to 97.5%
Custom Next.js 16 83.3% 68.8% to 97.5%
Docusaurus 12 68.4% 50.0% to 96.2%
Fern 8 76.7% 66.2% to 81.2%
Astro Starlight 6 68.8% 42.5% to 88.8%
GitBook 4 86.9% 76.2% to 95.0%
Hugo 2 82.5% 82.5% to 82.5%
ReadMe 2 81.9% 80.0% to 83.8%
Fumadocs 2 79.4% 78.8% to 80.0%
Gatsby 2 65% 60.0% to 70.0%
Custom Vike and React 1 82.5% SurrealDB
Custom Astro 1 81.2% Runware
Custom in-house 1 78.8% Stripe
Redocly 1 76.2% Mailgun
Custom SvelteKit 1 68.8% Amp
Custom Deno Deploy site 1 67.5% Deno
Antora and Asciidoctor 1 57.5% Neo4j
Jekyll 1 56.2% DuckDB
Nextra 1 53.8% Arcade.dev
MkDocs Material 1 51.2% Orkes
Quarto 1 23.8% TileDB

A platform with one tested product shows that product’s result and nothing more. Averages become meaningful as the count grows.

Put another docs site through the battery.

Nominate a docs site