Agent Readiness Auditor

Name: Agent Readiness Auditor
Author: asish-singh

by asish-singhversion 0.5.0i★ –iRuns on your computer

Registry nameio.github.asish-singh/agent-readiness-auditor

Install

Overview

Description from the maker (original English)

Audits any website for AI agent readiness and safety, scoring prompt injection risk and llms.txt.

PopNx MCP’s three pillars

These three are never merged into one score: opinions, measurements and popularity are different things.

User voiceiNo reviews yetReviews open together with accounts in a later step
Our basic checksi1 of 1 passedChecked from public information; code not scanned yet
UsageiVery new↓ 74/wk · ★ –

What you can do with it

We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.

Connection: Runs on your computer
Transport: stdio
Category: Search & AI memory

Install

Pick the AI app you use, then copy the settings.

Before you install

Basic checks: 1 of 1 passed
Runs on your computer with your user’s permissions (files, network, apps).
The registry lists no key to fill in.

You need

npmNode.js installed (for the npx command)

Which app do you use?

ClaudeClaude CodeCursorVS CodeChatGPT

Pick your app above. We remember it, so next time the button says “Install in …”.

Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json

{
  "mcpServers": {
    "agent-readiness-auditor": {
      "command": "npx",
      "args": [
        "-y",
        "agent-readiness-auditor",
        "mcp"
      ]
    }
  }
}

Run this in your terminal

claude mcp add agent-readiness-auditor -- npx -y agent-readiness-auditor mcp

Open in Cursor

Or copy the settings yourself

Paste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json

{
  "mcpServers": {
    "agent-readiness-auditor": {
      "command": "npx",
      "args": [
        "-y",
        "agent-readiness-auditor",
        "mcp"
      ]
    }
  }
}

Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.

Open in VS Code

Or copy the settings yourself

Paste into .vscode/mcp.json in your project

{
  "servers": {
    "agent-readiness-auditor": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "agent-readiness-auditor",
        "mcp"
      ]
    }
  }
}

Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.

This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.

Your choice stays on this device in a cookie; no account needed.

These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.

What it can do

Runs on your computer; the tool list appears in your AI app after you install it.

Basic checksi

We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 8, 2026

✓ Pass
Maker matches the repo
The registry name was verified with the GitHub account asish-singh, which owns the repo
… Not checked yet
Still maintained
We have not read this repo from GitHub yet
… Not checked yet
Has a licence
We have not read this repo from GitHub yet
– Not applicable
Endpoint responds
This one runs on your computer; there is no internet endpoint
– Not applicable
Sign-in with OAuth
Runs on your computer and uses local settings instead
… Not checked yet
No hidden instructions in tool descriptions
Runs on your computer; we can only read tool lists from internet endpoints

Compared with others in Search & AI memoryi

More stars than 0% of the other 5690 servers in this category
Published to the registry more recently than 59% of the other 5690 servers in this category
Basic checks equal to or better than 100% of the other 5690 servers in this category

Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.

Common questions

Is Agent Readiness Auditor safe?

It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.

What keys or codes does it need?

The registry does not list a required key, though some servers ask you to sign in on first use.

Does it run on my computer or in the cloud?

It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).

Is it still maintained?

The GitHub repo last changed –, and the latest version (0.5.0) reached the registry 27 days ago.

Which apps can use it?

Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.

What can I use instead?

In the same category: Context7, Codebase Memory, Exa

Alternatives

Servers in the same category from other makers, sorted by stars.

Reading a card:↓ Downloads per weeki★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi

Context7

Feeds AI the latest docs for libraries and frameworks before it answers, so code it writes does not use outdated, removed functions.

by upstashWidely used↓ 972.7K/wk★ 64.2KUpdated 2 days ago

Computer or cloud

Install

Codebase Memory

Codebase knowledge graph for AI agents — 162 languages, sub-ms queries, 99% fewer tokens.

by DeusDataWidely used↓ 18K/wk★ 48.1KUpdated 22 days ago

Runs on your computer

Install

Exa

Connect AI agents to Exa for web search, content fetching, and multi-step research.

by exa-labsGrowing★ 5.2KUpdated 2 months ago

Runs in the cloud✓ exa.ai

Install

README from the maker

Open the README

Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.

Agent Readiness Auditor

A command-line tool that scans a website and scores how well it works with, and how well it defends against, automated AI agents.

What problem this solves

AI assistants increasingly visit websites on a person's behalf: reading pages, following instructions found on them, and taking actions. This creates two practical issues for site owners:

Safety. A page can contain text that is hidden from human visitors but still read by an AI agent. That hidden text can carry instructions designed to hijack the agent. This is known as indirect prompt injection.
Readability for machines. Sites are usually built for human eyes. Agents do better when a site also publishes machine-readable signals about its content and its rules.

Standard SEO tools measure how well a site works for search engines. This tool measures something different: how well a site works for AI agents, and whether it is safe for them to read.

Research built with this tool

The State of the Agentic Web, 2026 audits 84 prominent sites across seven categories with this tool, and its second edition reruns the study with the deeper nine check audit. Among the findings, no audited site carries hidden prompt injection content even under attribute level and multi page scanning, only 13% publish the FAQ structured data AI answers quote, no site advertises an MCP endpoint, and nearly one in five prominent sites refuses automated visitors entirely, including several AI companies. The data and method are published for anyone to reproduce.

What it checks

The tool fetches the landing page, robots.txt, llms.txt, the sitemap, and the AI Catalog, then crawls a few key pages and runs nine checks, producing a score from 0 to 100 and a letter grade from A to F. Safety is weighted highest on purpose, because a readable site that can hijack an agent is worse than one that is simply hard to read.

Check	Points	What it looks for
Hidden prompt-injection text	40	Agent-hijacking phrases in content humans cannot see, across every crawled page. Detects CSS hiding (`display:none`, `opacity:0`, off-screen positioning), zero-width unicode obfuscation, `alt`/`title`/`aria-label` attribute payloads, and HTML comments.
`llms.txt` file	12	A published file that gives agents a curated map of the site. See llmstxt.org.
`robots.txt` stance on AI crawlers	12	Explicit allow or disallow rules for AI crawlers such as GPTBot and ClaudeBot.
Structured data (JSON-LD)	12	Machine-readable data that lets agents understand page content directly.
Accountability links	9	Reachable contact, privacy, terms, or about links.
Sitemap	5	A reachable /sitemap.xml so crawlers can discover the whole site.
Answerability	5	FAQ, QA, or HowTo structured data, the content shapes AI answers quote most readily.
Meta robots	5	No accidental `noindex`/`noai` directives hiding the site from indexes.
MCP signal	0	Informational only: whether the site publishes an AI Catalog at `/.well-known/ai-catalog.json` (the discovery document adopted by MCP Server Cards, SEP-2127) and which MCP server cards and A2A agent cards it lists, falling back to mentions in llms.txt or robots.txt. A malformed catalog is flagged with a fix, still unscored.

Example output

$ npx agent-readiness-auditor example.com

Agent readiness audit for https://example.com

✅ No hidden prompt-injection payloads (40/40) ⚠️ llms.txt present (0/12) ⚠️ robots.txt addresses AI crawlers (0/12) ⚠️ Machine-readable structured data (0/12) ⚠️ Accountability surface present (0/9) ⚠️ Sitemap present (0/5) ⚠️ Content structured for AI answers (0/5) ✅ No accidental index blocking (5/5) ✅ Agent endpoint advertised (informational) (0/0)

Score: 45/100 (45%) Grade D

Each line shows the check result, its score, and (when a check does not fully pass) a suggested fix.

Quick start

If you have Node.js 18 or newer installed, you can run the tool in one line without installing anything:

npx agent-readiness-auditor example.com

npx agent-readiness-auditor example.com --json

Use a bare domain (example.com) or a full URL (https://example.com). The --json flag prints machine-readable output for use in scripts.

Auditing many sites at once

Batch mode reads a text file with one URL per line (lines starting with # are ignored) and audits them a few at a time.

agent-audit --batch sites.txt          # a report per site

agent-audit --batch sites.txt --csv # one CSV row per site agent-audit --batch sites.txt --json # structured results

The CSV has a column for each check, which makes it easy to open in a spreadsheet or feed into an analysis.

Using it in GitHub Actions

The repo doubles as a GitHub Action, so any project can audit its own site on every push or on a schedule and fail the build if a hard safety problem appears.

steps:

uses: actions/setup-node@v4 with: node-version: 20
uses: asish-singh/agent-readiness-auditor@v0.4.1 with: url: example.com fail-on: safety # or "never" to report without failing

An unreachable site produces a warning, not a failure, since bot protection on CI runners is an infrastructure issue rather than a safety one. This repo uses the action on itself every Monday in audit-site.yml.

Using it from an AI assistant (MCP)

The auditor ships with a server for the Model Context Protocol, the standard that lets AI assistants use external tools, and is listed on the official MCP registry. One of this tool's own findings is that zero of 69 prominent sites advertise an agent endpoint, so the auditor holds itself to the standard it measures. Once connected, you can simply ask your assistant to audit a site for you.

For Claude Code, one command connects it.

claude mcp add agent-readiness-auditor -- npx -y agent-readiness-auditor mcp

For Claude Desktop, add this to the mcpServers section of the configuration file.

"agent-readiness-auditor": { "command": "npx", "args": ["-y", "agent-readiness-auditor", "mcp"] } }

The server exposes one tool, audit_site, which takes a URL and returns the same scores and findings as the command line.

Running from source

To work on the code or run it from a local copy:

git clone https://github.com/asish-singh/agent-readiness-auditor.git

cd agent-readiness-auditor npm install npm run audit -- example.com # human-readable report npm run audit -- example.com --json # JSON output

The -- in the command passes the URL to the tool rather than to npm.

To install a global agent-audit command from your local copy:

npm run build   # compile TypeScript into dist/

npm link # register the global command

You can then run agent-audit example.com from any folder. To remove it later, run npm unlink -g agent-readiness-auditor.

Installing from GitHub Packages

The package is also published to GitHub Packages as a scoped mirror, @asish-singh/agent-readiness-auditor. For most people the npm install above is simpler. Use GitHub Packages only if your organization standardizes on it, since it requires authentication even for public packages.

To install from it, create a GitHub personal access token with the read:packages scope, then point the scope at the GitHub registry:

echo "@asish-singh:registry=https://npm.pkg.github.com" >> .npmrc

echo "//npm.pkg.github.com/:_authToken=YOUR_GITHUB_TOKEN" >> .npmrc npm install @asish-singh/agent-readiness-auditor

Exit codes

The tool sets its exit code so it can be used in automated pipelines:

0: the audit ran and found no hard safety failure.
2: a hard safety failure was found (for example, hidden prompt-injection text). Use this to fail a build.
1: the tool could not complete the audit (for example, the site was unreachable).

How the code is organized

Each check lives in its own file under src/checks/ and returns a structured result. All checks are registered in a single list in src/audit.ts, and the total score is derived from that list, so adding a new check does not require changing the scoring logic. Architecture decisions are recorded in docs/adr/, product decisions in docs/decisions/, planned work in ROADMAP.md, and the go-to-market plan in LAUNCH.md.

Background

This tool grew out of the Agentic Web Governance Pack, a set of guidelines for how websites should behave toward AI agents. This project turns several of those guidelines into checks that can be run and measured.

License

MIT. See LICENSE.

Agent Readiness Auditor

Overview

PopNx MCP’s three pillars

What you can do with it

Install

Before you install

What it can do

Basic checksi

Maker matches the repo

Still maintained

Has a licence

Endpoint responds

Sign-in with OAuth

No hidden instructions in tool descriptions

Compared with others in Search & AI memoryi

Common questions

Alternatives

Context7

Codebase Memory

Exa

README from the maker

Agent Readiness Auditor

What problem this solves

Research built with this tool

What it checks

Example output

Quick start

Auditing many sites at once

Using it in GitHub Actions

Using it from an AI assistant (MCP)

Running from source

Installing from GitHub Packages

Exit codes

How the code is organized

Background

License