About
A practical guide to agent-ready software.
Agent Surface explains how to make a website, app, API or CLI work with AI agents, and how to build agents of your own. Daniel Howells writes and maintains it.
What it covers
The guide has two parts. Make your product agent-ready covers discovery, context files, APIs, tools, CLIs, MCP servers, authentication, errors, retrieval, agentic UI and agentic commerce. Build agents is a current inventory of the frameworks, protocols and tools for teams building agents of their own.
What you can use
Everything here is free to read and reuse under the MIT licence: the docs, the Surface skill for coding agents, the starter templates, and the scorecard used to audit a codebase.
Agents can read the site too. Every page is available as Markdown, the whole site is indexed in llms.txt, and an MCP server at /mcp searches and reads the docs. The server is listed in the official MCP Registry as dev.agentsurface/docs. None of it needs an account or a key.
How it's kept current
Pages about fast-moving tools carry a last-verified date, and a page that goes 120 days without a check is flagged to be verified again against primary sources. Model names in examples come from a single allowlist, so an outdated name fails the build.
The site itself is tested with the public agent-readiness checkers from Google Lighthouse, Cloudflare, Ora and Vercel, and the scanner guide explains what each one measures.
Who makes it
Agent Surface is made by Daniel Howells. More about him at danielhowells.com. The source is public on GitHub.
Corrections
If something is wrong or out of date, get in touch or open an issue on GitHub with the page and a link to the source that shows the correct answer.