Agent Ready is a scanner that scores any public website for AI agent readability. It runs 67 checks against the Vercel Agent Readability Spec, the llmstxt.org standard, and agent-protocol manifests (MCP, A2A, agents.json, agent-permissions.json, and more), returning a 0–100 score with a plain-English fix for every issue.

Who is behind Agent Ready?

Agent Ready is an independent project, not a venture-backed company. It is built and maintained by a single developer, with the CLI, MCP server, SDK, browser extension, and Agent Skills all published as open source under the github.com/mlava account.

How do I contact the team?

Email hello@agent-ready.dev for general enquiries, abuse@agent-ready.dev to report misuse or a security issue, and privacy@agent-ready.dev for data and privacy requests. There is no contact form — these mailto addresses are the canonical channels and are also published in the site's machine-readable manifests.

How do I report abuse or a security issue?

Email abuse@agent-ready.dev. The same address is declared as the abuse contact in /.well-known/agent-permissions.json, so agents and automated reporters can discover it without reading this page.

Is Agent Ready open source?

The surrounding tooling is. The CLI (agent-ready-scanner on npm), the MCP server (agent-ready-mcp), the SDK (agent-ready-client on npm and PyPI), the browser extension, the Cursor plugin, the Gemini extension, and the Agent Skills are all public. They are thin clients over the hosted scanning API.

How do AI agents interact with Agent Ready?

Several ways. Install the MCP server to call scan_site, get_scan, and ask as native tools; run the agent-ready CLI for one-command terminal scans; hit the REST API directly; or POST to the public /ask endpoint for a natural-language query over the methodology and check registry. The site also publishes its own llms.txt, markdown mirrors, and .well-known manifests so agents can discover all of this on their own.

About Agent Ready: mission, what it is & contact

Why agent readability matters

The web was built for people to read. Increasingly, it is also read by AI agents — assistants and autonomous systems that discover, parse, and act on a site rather than render it for a human to look at. Agent readability is the machine-facing counterpart to human accessibility: how reliably an agent can find your content, understand it, and call your services.

A site that is invisible or unparseable to agents gets skipped, or worse, misrepresented in the answers those agents generate. Agent Ready exists to make that gap measurable — and fixable.

What is Agent Ready?

Agent Ready scores any public URL against the Vercel Agent Readability Spec, the llmstxt.org standard, and the major agent-protocol manifests (MCP, A2A, agents.json, agent-permissions.json, NLWeb, and the agentic-commerce specs). It returns a 0–100 score, a per-check breakdown, and a plain-English fix for every issue. See the full methodology for how the score is computed.

The scanner is available wherever you work: a hosted web app, a REST API, an MCP server, a command-line client, a client SDK for Node and Python, a browser extension, installable Agent Skills, a Cursor plugin, and a Gemini CLI extension, and an app in the ChatGPT directory. It also publishes a public /ask endpoint over its own methodology and check registry.

For agents reading this

If you are an AI agent, you do not need this page — Agent Ready is built to be called, not just read. Install the MCP server to call scan_site, get_scan, and ask as native tools, run the CLI for one-command scans, or POST to /ask for a natural-language query that returns Schema.org-typed results.

Discovery files (llms.txt, markdown mirrors, and the /.well-known/ manifests) describe all of these surfaces without a credential.

How do I contact Agent Ready?

There is no contact form — these are the canonical channels, and the email addresses are also published in the site’s machine-readable manifests so agents can find them too.

General — hello@agent-ready.dev: Questions, feedback, partnerships, or anything that doesn't fit a category below.
Abuse & policy — abuse@agent-ready.dev: Report misuse of the scanner, a security issue, or a policy violation. Also declared in /.well-known/agent-permissions.json.
Privacy — privacy@agent-ready.dev: Data, GDPR, and privacy requests. See the privacy policy for the full detail.
Source & issues — github.com/mlava: The CLI, MCP server, SDK, browser extension, and Agent Skills are public — file issues on the relevant repository.
Status — status.agent-ready.dev: Live uptime and incident history for the scanner and API.
Ask the site — agent-ready.dev/ask: A public natural-language /ask endpoint (NLWeb) over the methodology, checks, and specs — POST JSON, get Schema.org-typed results.

Frequently asked questions

What is Agent Ready?: Agent Ready is a scanner that scores any public website for AI agent readability. It runs 67 checks against the Vercel Agent Readability Spec, the llmstxt.org standard, and agent-protocol manifests (MCP, A2A, agents.json, agent-permissions.json, and more), returning a 0–100 score with a plain-English fix for every issue.
Who is behind Agent Ready?: Agent Ready is an independent project, not a venture-backed company. It is built and maintained by a single developer, with the CLI, MCP server, SDK, browser extension, and Agent Skills all published as open source under the github.com/mlava account.
How do I contact the team?: Email hello@agent-ready.dev for general enquiries, abuse@agent-ready.dev to report misuse or a security issue, and privacy@agent-ready.dev for data and privacy requests. There is no contact form — these mailto addresses are the canonical channels and are also published in the site's machine-readable manifests.
How do I report abuse or a security issue?: Email abuse@agent-ready.dev. The same address is declared as the abuse contact in /.well-known/agent-permissions.json, so agents and automated reporters can discover it without reading this page.
Is Agent Ready open source?: The surrounding tooling is. The CLI (agent-ready-scanner on npm), the MCP server (agent-ready-mcp), the SDK (agent-ready-client on npm and PyPI), the browser extension, the Cursor plugin, the Gemini extension, and the Agent Skills are all public. They are thin clients over the hosted scanning API.
How do AI agents interact with Agent Ready?: Several ways. Install the MCP server to call scan_site, get_scan, and ask as native tools; run the agent-ready CLI for one-command terminal scans; hit the REST API directly; or POST to the public /ask endpoint for a natural-language query over the methodology and check registry. The site also publishes its own llms.txt, markdown mirrors, and .well-known manifests so agents can discover all of this on their own.