Agents already use docs as product instructions
Coding agents, AI search products, and automated support flows read docs long before a human asks for help. If the docs are hard to fetch or parse, the product feels harder to use.
Documentation benchmark
The docs benchmark for AI-readable documentation.
Run a documentation benchmark on any public docs site and see whether an agent can find the right page, read clean text, and follow the instructions without guesswork across API docs, developer docs, and help-center docs.
Use it as a public docs leaderboard, an API docs benchmark, or a way to compare how different documentation sites hold up when an AI agent tries to use them.
227 tracked docs7 benchmark categories72 average benchmark scoreAFDocs-informed scoring
Try one of these public docs URLs:
We run the crawl here first, then open the full report with scores, failed checks, and next fixes.
Leaderboard
Browse public benchmark reports, compare category scores, and open the full breakdown for each docs site.
Categories7 tracked segments
Showing 25 of 227 matching docs.
Why it matters
Docs quality now affects AI usability.
This is not a vanity score. It measures whether an agent can discover, read, and act on your documentation.
Docs quality now affects activation
Agent-readable docs shape implementation speed, support load, and whether a product feels trustworthy during evaluation.
Benchmarks make docs quality comparable
A shared benchmark gives teams a concrete way to compare docs quality, spot weak areas, and track improvements over time.
Methodology
How the benchmark scores docs.
This scoring model is informed by AFDocs and adapted into a public benchmark teams can inspect, compare, and rerun.
Checks for llms.txt, sitemaps, and a clear public docs entry point.
- LLMS TXT existsAn llms.txt file is published at a predictable docs path.
- LLMS TXT is structuredThe llms.txt file follows a format agents can parse quickly.
- Docs entry point is obviousThe primary docs home is easy to identify.
- Docs sitemap existsThe crawl map is present and reachable.
- Key docs links resolveThe sampled entry links resolve without dead ends.
- Public indexing path is stableAgents can stay on a stable public path while crawling the docs.
Built by DocsAlot
Need help improving the score?
DocsAlot helps teams improve help centers, developer docs, API docs, and CLI docs so they are easier for humans to use and easier for agents to read.
Resources
Learn how to improve a docs benchmark score.
These pages help teams move from a benchmark report to concrete improvements in docs structure, AI discoverability, and developer onboarding.