Docs for Agents

Docs platform

Docusaurus docs and AI agent readiness

12 of the 97 products tested serve their documentation on Docusaurus.

Scores run from 50.0% to 96.2%, averaging 68.4%. Every Docusaurus product tested passes llms.txt. The AI Agent Readiness Test scores the documentation, never the platform, so these numbers describe what these 12 teams published rather than what Docusaurus can do.

Scores on Docusaurus

Product Category Grade Score Reading Surface llmsfull.mdMCPAI
MotherDuck Databases & analytics A 96.2% 27/30 50/50
Convex Databases & analytics A 95.0% 26/30 50/50
Mapbox Geospatial & maps B- 80.0% 24/30 40/50
Mastra Agent frameworks & memory B- 80.0% 24/30 40/50
Airbyte Data pipelines & documents D+ 67.5% 24/30 30/50
Speechmatics Voice & speech D 65.0% 22/30 30/50
Ory Identity & auth D- 62.5% 20/30 30/50
Zilliz Cloud Search & retrieval D- 60.0% 18/30 30/50
Overture Maps Geospatial & maps F 58.8% 27/30 20/50
Hermes Agent Coding agents F 53.8% 23/30 20/50
Fused Geospatial & maps F 52.5% 22/30 20/50
Vectara Search & retrieval F 50.0% 20/30 20/50

Agent surface on Docusaurus

  • llms.txt 12 of 12
  • llms-full.txt 6 of 12
  • Markdown mirror 9 of 12
  • MCP server 4 of 12
  • Docs AI 7 of 12

Recorded as Docusaurus, Docusaurus 3.6.3, Docusaurus 3.9.2, Docusaurus 3.10.0, Docusaurus 3.10.1, Docusaurus 3.10.2 during testing. A sample of 12 shows what these teams shipped and does not establish a platform default.

All docs platforms · Every score in one table

Put another docs site through the battery.

Nominate a docs site