The scoreboard

Every AI Agent Readiness Score

One row per product, ranked by AI Agent Readiness Score.

The AI Agent Readiness Score rates a product’s public documentation out of 80 points. Fifteen reading votes from a three-model panel are worth 30, and five checks on the agent-facing surface are worth 50. The table below ranks every product tested so far and shows which of the five surface checks each one passes.

16 products tested

AI Agent Readiness Score for every product tested, ranked high to low.
Product Grade Score Reading Surface llmsfull.mdMCPAI Platform
Felt A 95.0% 26/30 50/50 pass pass pass pass pass GitBook
Browserbase A- 92.5% 24/30 50/50 pass pass pass pass pass Mintlify
Resend B+ 88.8% 21/30 50/50 pass pass pass pass pass Mintlify
E2B B+ 87.5% 20/30 50/50 pass pass pass pass pass Mintlify
LangChain B+ 87.5% 20/30 50/50 pass pass pass pass pass Mintlify
CARTO B- 82.5% 16/30 50/50 pass pass pass pass pass GitBook
Convex B- 82.5% 26/30 40/50 pass pass pass fail pass Docusaurus 3.10.1
Firecrawl B- 81.2% 25/30 40/50 pass pass pass pass fail Mintlify
Composio B- 80.0% 24/30 40/50 pass pass pass fail pass Fumadocs on Next.js
Stripe C+ 78.8% 23/30 40/50 pass fail pass pass pass custom in-house
Mailgun C 76.3% 21/30 40/50 pass fail pass pass pass Redocly
Railway C 73.8% 19/30 40/50 pass pass pass pass fail Custom Next.js
Airbyte D+ 67.5% 24/30 30/50 pass fail pass fail pass Docusaurus 3.10.1
Daytona D- 62.5% 20/30 30/50 pass pass pass fail fail Astro Starlight
DuckDB F 56.2% 25/30 20/50 pass fail pass fail fail Jekyll 4.4.1
Arcade.dev F 53.8% 23/30 20/50 pass fail pass fail fail Nextra

Agent surface adoption

  • llms.txt 16 of 16 products
  • llms-full.txt 11 of 16 products
  • Markdown mirror 16 of 16 products
  • MCP server 10 of 16 products
  • Docs AI 11 of 16 products

Put another docs site through the battery.

Nominate a docs site