WP Manifestindependent plugin directory
manifest / ai / ia-agent-ready-website

IA Agent Ready Website

WordPress plugin that prepares your site for the era of AI agents — Link headers (RFC 8288), robots.txt AI bots & Content Signals, MCP Server Card, Agent Skills Index, OAuth/OIDC, Markdown Negotiation, WebMCP and more.

by Supersonique Studio · github.com/mikecastrodemaria/ia-agent-ready-website · website

★ 1stars
0forks

Install

No release zip yet. The repository archive installs, but the folder name will carry the branch suffix and updates will not flow:

wp plugin install https://github.com/mikecastrodemaria/ia-agent-ready-website/archive/refs/heads/main.zip
Readme github.com/mikecastrodemaria/ia-agent-ready-website ↗

WordPress plugin that prepares your site for the era of AI agents — Link headers (RFC 8288), robots.txt AI bots & Content Signals, MCP Server Card, Agent Skills Index, OAuth/OIDC, Markdown Negotiation, WebMCP and more.

Why this plugin?

Cloudflare's isitagentready.com audit rates a website's "Agent-Readiness" out of 100. Without any preparation, a vanilla WordPress site scores around 25/100 (Level 1 – Basic Web Presence).

This plugin fixes almost every issue automatically from the WordPress admin, by generating the files and headers expected by modern AI agents.

Category Before After
Discoverability 67 (2/3) 100 (3/3)
Content (Markdown) 0 (0/1) 100 (1/1)
Bot Access Control 50 (1/2) 100 (2/2)
API/Auth/MCP & Skills 0 (0/6) 100 (6/6)
Overall score 25 92–100 (Level 5 – Agent-Native)

Serve mode: Static (default) vs Dynamic

Since v1.3.0, the plugin offers two modes:

📁 Static (default) ⚡ Dynamic
How Writes real JSON files in ABSPATH/.well-known/ WordPress rewrite rules + PHP
Performance ~5 ms (Apache static) ~100 ms (WP boot)
Cache-friendly ✅ Cloudflare caches the 200 ⚠️ Risk of caching the initial 404
Survives WP outages ✅ ❌
Requires .well-known/ writable .htaccess routing to index.php

Dynamic mode remains active as a fallback: if a static file is missing, Apache routes the request to WordPress which serves the content via PHP.

Features

① Link Response Headers (RFC 8288)

Link: HTTP headers + <link> tags in <head> for automatic discovery:

Link: </.well-known/api-catalog>; rel="api-catalog"
Link: </.well-known/agent-skills/index.json>; rel="agent-skills"

② robots.txt – AI bot rules

Directives for 12 AI crawlers via the WordPress robots_txt filter:

User-agent: GPTBot
Disallow:

User-agent: Claude-Web
Disallow:

User-agent: anthropic-ai
Disallow:
...

③ robots.txt – Content Signals

contentsignals.org standard:

Content-Signal: ai-train=no, search=yes, ai-input=no

④ MCP Server Card (SEP-2127)

/.well-known/mcp/server-card.json:

{
  "$schema": "https://spec.modelcontextprotocol.io/schemas/server-card/v1.json",
  "serverInfo": { "name": "My Site", "version": "1.0.0" },
  "transport": { "type": "http", "url": "https://mysite.com/mcp" },
  "capabilities": { "tools": true, "resources": false, "prompts": false }
}

⑤ Agent Skills Index (RFC v0.2.0)

/.well-known/agent-skills/index.json with search and sitemap skills. Extensible via supup_agent_skills.

⑥ API Catalog (RFC 9727)

/.well-known/api-catalog formatted as application/linkset+json.

⑦ OAuth Protected Resource (RFC 9728)

/.well-known/oauth-protected-resource:

{
  "resource": "https://mysite.com",
  "authorization_servers": ["https://mysite.com"],
  "scopes_supported": ["read", "write"],
  "bearer_methods_supported": ["header"]
}

⑧ OAuth/OIDC Discovery (RFC 8414 / OpenID Connect)

/.well-known/openid-configuration + /.well-known/oauth-authorization-server with issuer, endpoints, grant_types, scopes, signing algs.

⑨ Web Bot Auth (RFC 9421)

/.well-known/http-message-signatures-directory formatted as application/http-message-signatures-directory+json.

⑩ Markdown Negotiation

When an agent sends Accept: text/markdown, the HTML is converted to Markdown on the fly:

curl -s -H "Accept: text/markdown" https://mysite.com/
# → headings, links, lists, bold, italic, code, images
# Content-Type: text/markdown; charset=utf-8
# Vary: Accept

⑪ WebMCP (navigator.modelContext)

Script injected into <head> (with retry) that registers tools via navigator.modelContext.provideContext(). Default tools: search, get_sitemap, get_homepage. Extensible via supup_webmcp_tools.

Installation

# Option A — direct copy
cp -r supersonique-agent-ready/ /path/to/wp-content/plugins/

# Option B — ZIP
cd .. && zip -r supersonique-agent-ready.zip supersonique-agent-ready/
# then upload via Plugins → Add New → Upload Plugin

Then: Plugins → Activate → Settings → 🤖 Agent Ready

On activation, the plugin:

  1. Creates ABSPATH/.well-known/ if needed
  2. Writes all JSON files (Static mode by default)
  3. Generates an .htaccess with the right Content-Types
  4. Registers rewrite rules as a fallback

🌩️ Cloudflare configuration

If your site is behind Cloudflare, a few Cloudflare-side settings are required for the plugin's signals to reach scanners and agents. Without them, the score may stay stuck around 70–80 even with the plugin correctly configured.

1. Transform Rule for the Link header (mandatory)

Cloudflare caches the homepage HTML and serves a cached version without the Link: headers sent by WordPress. Fix: inject the header at the edge.

Cloudflare dashboard → your domain → Rules → Transform Rules → Modify Response Header → Create rule

Field Value
Rule name Agent-Ready Link header
When incoming requests match Hostname equals your-domain.com
Action Set static
Header name Link
Value </.well-known/api-catalog>; rel="api-catalog", </.well-known/agent-skills/index.json>; rel="agent-skills"

This rule applies to every response (cached or not) — no cache purge needed.

2. Markdown for Agents (recommended)

Cloudflare provides Markdown negotiation natively, faster than the PHP converter and with better rendering:

Cloudflare dashboard → AI → AI Crawl Control → Markdown for Agents → ON

⚠️ If you enable Cloudflare's Markdown for Agents: disable ④b Markdown Negotiation in the plugin admin to avoid double conversion.

If the feature isn't available on your Cloudflare plan:

  • Keep the plugin handler enabled (④b)
  • Add a Cache Rule: match (http.host eq "your-domain.com" and http.request.headers["accept"][0] contains "text/markdown") → action Bypass cache

3. Purge Everything after each change

Caching → Configuration → Purge Everything

Required after:

  • First activation of the plugin
  • Adding/modifying Transform Rules
  • Switching Static ↔ Dynamic mode
  • Every settings save if you want agents to see the new .well-known/ files right away

Without a purge, the 404s Cloudflare cached on the old endpoints keep being served for hours.

4. Disable APO

Cloudflare APO aggressively caches HTML and strips some headers (including Link:).

Speed → Optimization → Content Optimization → Cloudflare APO → OFF

If you want to keep it, the Transform Rule from §1 works around the problem.

5. Page Rules or Cache Rules for .well-known/

Optional but clean — ensures the static JSON files are cached correctly:

Caching → Cache Rules → Create rule

Field Value
Name .well-known JSON cache
Match URI Path starts with /.well-known/
Action Cache eligibility → Eligible for cache
Edge TTL 2 hours (override origin)
Browser TTL 1 hour

6. Verification

# Link header should appear
curl -sI https://your-domain.com/ | grep -i link

# All .well-known endpoints should return 200
curl -sI https://your-domain.com/.well-known/api-catalog
curl -sI https://your-domain.com/.well-known/oauth-protected-resource
curl -sI https://your-domain.com/.well-known/openid-configuration
curl -sI https://your-domain.com/.well-known/http-message-signatures-directory
curl -sI https://your-domain.com/.well-known/mcp/server-card.json
curl -sI https://your-domain.com/.well-known/agent-skills/index.json

# Markdown negotiation
curl -sI -H "Accept: text/markdown" https://your-domain.com/ | grep -i content-type

If everything is ✅ and the scanner still fails, it's a cache issue: Purge Everything and wait 60 seconds before re-testing.

Cloudflare checklist

  • [ ] Transform Rule "Agent-Ready Link header" deployed
  • [ ] Markdown for Agents enabled (or Cache Rule bypass for Accept: text/markdown)
  • [ ] APO disabled (or Transform Rule in place)
  • [ ] Purge Everything done after installing the plugin
  • [ ] (Optional) Cache Rule for .well-known/*
  • [ ] curl verification OK on the 6 endpoints

Configuration

All settings live in WP Admin → Settings → 🤖 Agent Ready.

After every change:

  1. Click 💾 Save settings — static files are regenerated automatically
  2. Purge the Cloudflare cache (Purge Everything)
  3. Purge the WordPress cache plugin (WP Rocket, W3TC, LiteSpeed…)
  4. Re-test on isitagentready.com

Quick verification

curl.exe -sI https://mysite.com/.well-known/oauth-protected-resource
curl.exe -sI https://mysite.com/.well-known/openid-configuration
curl.exe -sI https://mysite.com/.well-known/http-message-signatures-directory
curl.exe -sI -H "Accept: text/markdown" https://mysite.com/

All should return 200 with the right Content-Type.

Extensibility

Add custom Agent Skills

add_filter( 'supup_agent_skills', function( array $skills ): array {
    $skills[] = [
        'name'        => 'contact',
        'type'        => 'action',
        'description' => 'Submit a project enquiry.',
        'url'         => home_url( '/contact' ),
        'inputSchema' => [
            'type'       => 'object',
            'properties' => [
                'name'    => [ 'type' => 'string' ],
                'email'   => [ 'type' => 'string', 'format' => 'email' ],
                'message' => [ 'type' => 'string' ],
            ],
            'required'   => [ 'email', 'message' ],
        ],
        'sha256'      => hash( 'sha256', home_url( '/contact' ) ),
    ];
    return $skills;
} );

Add custom WebMCP tools

add_filter( 'supup_webmcp_tools', function( array $tools ): array {
    $tools[] = [
        'name'        => 'list_latest_posts',
        'description' => 'List the 10 most recent blog posts.',
        'inputSchema' => [ 'type' => 'object', 'properties' => (object) [] ],
    ];
    return $tools;
} );

Supported skill types

Type Description
search / query Search inside content
navigation Navigation to resources
action Triggers an action (form, API call)
feed Data feed (RSS, JSON feed…)

Internationalization

Source code and admin UI are in English. A French translation ships in /languages/supersonique-agent-ready-fr_FR.po (compile to .mo with msgfmt or any WordPress translation tool).

To add another language:

cd languages/
# Update the POT file by re-running WP-CLI:
wp i18n make-pot ../ supersonique-agent-ready.pot
# Create a new locale, e.g. German:
msginit --locale=de_DE --input=supersonique-agent-ready.pot
# Translate de_DE.po, then:
msgfmt supersonique-agent-ready-de_DE.po -o supersonique-agent-ready-de_DE.mo

Standards implemented

Standard Description URL
RFC 8288 Web Linking – Link headers rfc-editor.org
RFC 8414 OAuth 2.0 Authorization Server Metadata rfc-editor.org
RFC 9421 HTTP Message Signatures rfc-editor.org
RFC 9727 API Catalog rfc-editor.org
RFC 9728 OAuth Protected Resource Metadata rfc-editor.org
OpenID Connect Discovery OIDC metadata openid.net
MCP SEP-2127 MCP Server Card GitHub PR
Agent Skills RFC v0.2.0 Agent Skills Discovery GitHub
Content Signals AI content preferences contentsignals.org
WebMCP navigator.modelContext webmachinelearning.github.io

FAQ

Static vs Dynamic mode? Static writes real files served by Apache (fast, cache-friendly). Dynamic uses WordPress + PHP on every request. If disk writes fail, Dynamic kicks in automatically.

Does the plugin work without Cloudflare? Yes. Every feature works on any Apache-backed WordPress host.

.well-known endpoints return 404 in Static mode? Check the admin: ABSPATH/.well-known/ must be writable. Otherwise chmod 755 public_html/ or switch to Dynamic mode.

.well-known endpoints return 404 in Dynamic mode? Settings → Permalinks → Save Changes to reload the rewrite rules.

Does the plugin modify the physical robots.txt? No. It uses the WordPress robots_txt filter — no file is touched.

Cloudflare doesn't show the Link headers on the homepage? Cloudflare caches the HTML without the header. See the 🌩️ Cloudflare configuration section above — the Transform Rule fixes it in 30 seconds.

The isitagentready.com scanner still shows 404 on .well-known/oauth-protected-resource while curl returns 200? Cloudflare cache. Caching → Configuration → Purge Everything, wait 60 s, re-scan.

Development

supersonique-agent-ready/
├── supersonique-agent-ready.php   # Main plugin file
├── readme.txt                     # WordPress.org format
├── README.md                      # This file
├── CHANGELOG.md                   # Version history
├── LICENSE                        # GPL-2.0-or-later
├── uninstall.php                  # Cleanup on uninstall
├── languages/                     # Translations
│   ├── supersonique-agent-ready.pot
│   └── supersonique-agent-ready-fr_FR.po
└── assets/
    ├── banner-1544x500.png
    └── icon-256x256.png

Hooks and filters

Hook Type Description
supup_agent_skills filter Modify the Agent Skills served
supup_webmcp_tools filter Modify the WebMCP tools
option_sar_settings filter Auto-merge defaults (internal — don't override)
update_option_sar_settings action Regenerates the static files

Internal helpers (Static mode)

Function Role
supup_static_root() Returns ABSPATH/.well-known/
supup_well_known_files() List of managed files
supup_build_payloads() Builds all the JSON payloads
supup_write_static_files() Writes the files + .htaccess
supup_delete_static_files() Removes files + empty subdirs

License

GPL-2.0-or-later © 2026 Supersonique Studio