| 1 |
<?php |
| 2 |
/** |
| 3 |
* MCP connection self-test. |
| 4 |
* |
| 5 |
* @package ThinkRank\Mcp |
| 6 |
*/ |
| 7 |
|
| 8 |
declare(strict_types=1); |
| 9 |
|
| 10 |
namespace ThinkRank\Mcp; |
| 11 |
|
| 12 |
use ThinkRank\Abilities\Abilities_Registrar; |
| 13 |
|
| 14 |
if ( ! defined( 'ABSPATH' ) ) { |
| 15 |
exit; // Exit if accessed directly. |
| 16 |
} |
| 17 |
|
| 18 |
/** |
| 19 |
* Exercises the MCP round trip the way an external client would and reports |
| 20 |
* *where* it broke, so an admin can tell a certificate problem from an |
| 21 |
* authentication problem from an ability-discovery problem without leaving the |
| 22 |
* MCP page (see #189). |
| 23 |
* |
| 24 |
* Four loopback checks, each against a surface a real client actually uses: |
| 25 |
* |
| 26 |
* 1. `endpoint` — the pretty URL the user pastes (/thinkrank/mcp), with the |
| 27 |
* connection token. Depends on rewrite rules, so it fails on |
| 28 |
* plain permalinks or an unflushed rule table. |
| 29 |
* 2. `fallback` — the always-on /wp-json/thinkrank/v1/mcp route. Works even |
| 30 |
* when rewrites do not, which is what separates "the whole |
| 31 |
* MCP surface is down" from "only the pretty URL is". |
| 32 |
* 3. `discovery` — the RFC 9728 / RFC 8414 metadata documents. |
| 33 |
* 4. `challenge` — an UNAUTHENTICATED call, which must answer 401 with a |
| 34 |
* WWW-Authenticate header. This is the only thing an |
| 35 |
* OAuth-only client (ChatGPT, claude.ai) has to go on: if |
| 36 |
* the challenge is missing it reports that the server does |
| 37 |
* not implement OAuth, no matter how healthy the rest is. |
| 38 |
* |
| 39 |
* Checks 3 and 4 exist because a token-pasting client can be perfectly happy |
| 40 |
* while every OAuth client is refused — the earlier version of this test only |
| 41 |
* exercised check 2 and so reported `ok` in exactly that situation. |
| 42 |
* |
| 43 |
* The staged result names the first failing step: `disabled` → `not_connected` |
| 44 |
* → `unreachable` → `tls` → `redirect` → `auth` → `no_tools` → `rewrite` → |
| 45 |
* `discovery` → `challenge` → `ok`. |
| 46 |
*/ |
| 47 |
final class Mcp_Self_Test { |
| 48 |
|
| 49 |
/** |
| 50 |
* User agents to replay the challenge probe with, to detect a host that |
| 51 |
* filters by User-Agent. These are the shapes real MCP backends send — |
| 52 |
* none of them is a browser, which is exactly what "block bad bots" rules |
| 53 |
* key on. A site that answers WordPress's own UA but 403s these is |
| 54 |
* unreachable for every AI client while looking perfectly healthy from |
| 55 |
* inside. |
| 56 |
* |
| 57 |
* DO NOT replace these with a descriptive agent such as |
| 58 |
* `ThinkRank-SelfTest/1.0`. An affected host allowlists a named agent and |
| 59 |
* keeps refusing `python-requests/…`, so the check would go green while |
| 60 |
* ChatGPT stays blocked — the exact false pass this test exists to catch. |
| 61 |
* SiteGround support has recommended that change; declining it is |
| 62 |
* deliberate. See #379. |
| 63 |
*/ |
| 64 |
private const CLIENT_USER_AGENTS = [ |
| 65 |
'python-requests/2.32.3', |
| 66 |
'node-fetch/3.3.2', |
| 67 |
]; |
| 68 |
|
| 69 |
/** |
| 70 |
* Where an affected site owner is sent for the workaround list. The plugin |
| 71 |
* cannot fix an edge block, so the failing check hands over the diagnostic |
| 72 |
* and the host-side options instead. |
| 73 |
*/ |
| 74 |
private const HOSTING_DOC_URL = 'https://thinkrank.ai/docs/mcp/hosting-compatibility/'; |
| 75 |
|
| 76 |
/** |
| 77 |
* Run the round-trip self-test. |
| 78 |
* |
| 79 |
* @return array<string, mixed> |
| 80 |
*/ |
| 81 |
public static function run(): array { |
| 82 |
$endpoint = Mcp_Pairing::site_endpoint(); |
| 83 |
$fallback = Mcp_Pairing::site_endpoint_fallback(); |
| 84 |
|
| 85 |
$result = [ |
| 86 |
'ok' => false, |
| 87 |
'stage' => '', |
| 88 |
'message' => '', |
| 89 |
'endpoint' => $endpoint, |
| 90 |
'endpoint_rest' => $fallback, |
| 91 |
'mcp_enabled' => Mcp_Manager::is_enabled(), |
| 92 |
'connected' => Mcp_Pairing::is_connected(), |
| 93 |
'http_status' => null, |
| 94 |
'redirected' => false, |
| 95 |
'authenticated' => false, |
| 96 |
'tools_count' => null, |
| 97 |
'checks' => [], |
| 98 |
// url => decoded metadata (or the raw body when it isn't JSON). |
| 99 |
'discovery_documents' => [], |
| 100 |
]; |
| 101 |
|
| 102 |
if ( ! $result['mcp_enabled'] ) { |
| 103 |
$result['stage'] = 'disabled'; |
| 104 |
$result['message'] = __( 'MCP access is turned off, so the endpoint refuses every request. Enable MCP access above and try again.', 'thinkrank' ); |
| 105 |
return $result; |
| 106 |
} |
| 107 |
|
| 108 |
if ( ! $result['connected'] ) { |
| 109 |
$result['stage'] = 'not_connected'; |
| 110 |
$result['message'] = __( 'No connection token exists yet. Click Connect to mint one, then run the test again.', 'thinkrank' ); |
| 111 |
return $result; |
| 112 |
} |
| 113 |
|
| 114 |
$token = Mcp_Pairing::site_token(); |
| 115 |
|
| 116 |
$pretty = self::probe_jsonrpc( $endpoint, $token ); |
| 117 |
$rest = self::probe_jsonrpc( $fallback, $token ); |
| 118 |
|
| 119 |
// Back-compat top-level fields describe the primary (pretty) endpoint, |
| 120 |
// falling back to the REST route when the pretty URL never answered. |
| 121 |
$primary = 'unreachable' === $pretty['stage'] ? $rest : $pretty; |
| 122 |
$result['http_status'] = $primary['status']; |
| 123 |
$result['redirected'] = 'redirect' === $primary['stage']; |
| 124 |
$result['authenticated'] = $primary['authenticated']; |
| 125 |
$result['tools_count'] = $primary['tools']; |
| 126 |
|
| 127 |
$result['checks'][] = self::check( 'endpoint', __( 'Connection URL', 'thinkrank' ), $pretty['stage'], $pretty['detail'] ); |
| 128 |
$result['checks'][] = self::check( 'fallback', __( 'REST fallback URL', 'thinkrank' ), $rest['stage'], $rest['detail'] ); |
| 129 |
|
| 130 |
$discovery = self::probe_discovery(); |
| 131 |
// The documents themselves, so support can read what the site actually |
| 132 |
// serves instead of asking the customer for screenshots. |
| 133 |
$result['discovery_documents'] = $discovery['documents']; |
| 134 |
$result['checks'][] = self::check( 'discovery', __( 'OAuth discovery', 'thinkrank' ), $discovery['stage'], $discovery['detail'] ); |
| 135 |
|
| 136 |
$challenge = self::probe_challenge( $endpoint, $fallback ); |
| 137 |
$result['checks'][] = self::check( 'challenge', __( 'OAuth challenge', 'thinkrank' ), $challenge['stage'], $challenge['detail'] ); |
| 138 |
|
| 139 |
// Only reported when it could actually run — claiming a pass we did |
| 140 |
// not measure is the failure mode this whole test exists to avoid. |
| 141 |
$user_agent = self::probe_user_agent( $endpoint ); |
| 142 |
if ( null !== $user_agent ) { |
| 143 |
$result['checks'][] = self::check( 'user_agent', __( 'Client access', 'thinkrank' ), $user_agent['stage'], $user_agent['detail'], $user_agent['doc_url'] ?? '' ); |
| 144 |
} |
| 145 |
|
| 146 |
// Locked-out clients. The loopback below can pass while a REMOTE client |
| 147 |
// is walled off by the failed-auth limiter — the exact state a connector |
| 148 |
// still holding a rotated-away token produces. Reported only when the |
| 149 |
// count is knowable (null under a persistent object cache). |
| 150 |
$lockouts = Mcp_Rate_Limiter::active_lockouts(); |
| 151 |
$result['locked_clients'] = $lockouts; |
| 152 |
if ( null !== $lockouts && $lockouts > 0 ) { |
| 153 |
$result['checks'][] = self::check( |
| 154 |
'lockouts', |
| 155 |
__( 'Client lockouts', 'thinkrank' ), |
| 156 |
'locked_clients', |
| 157 |
sprintf( |
| 158 |
/* translators: %d: number of currently locked-out clients. */ |
| 159 |
_n( |
| 160 |
'%d client is currently locked out after repeated failed authentications — typically a connector still holding a rotated-away token. Remove and re-add the connector in the AI client; the lockout clears itself within 15 minutes of the retries stopping.', |
| 161 |
'%d clients are currently locked out after repeated failed authentications — typically connectors still holding a rotated-away token. Remove and re-add the connector in the AI client; lockouts clear within 15 minutes of the retries stopping.', |
| 162 |
$lockouts, |
| 163 |
'thinkrank' |
| 164 |
), |
| 165 |
$lockouts |
| 166 |
) |
| 167 |
); |
| 168 |
} |
| 169 |
|
| 170 |
// The pretty URL failing while the fallback works is its own finding: |
| 171 |
// the site is usable, but only via the REST URL. |
| 172 |
if ( 'ok' !== $pretty['stage'] && 'ok' === $rest['stage'] ) { |
| 173 |
$result['stage'] = 'rewrite'; |
| 174 |
$result['message'] = sprintf( |
| 175 |
/* translators: 1: pretty MCP endpoint URL, 2: REST fallback URL. */ |
| 176 |
__( 'The connection URL %1$s did not answer, but the REST fallback %2$s works. Re-save Settings → Permalinks to rebuild the rewrite rules; until then, give your AI client the fallback URL.', 'thinkrank' ), |
| 177 |
$endpoint, |
| 178 |
$fallback |
| 179 |
); |
| 180 |
return $result; |
| 181 |
} |
| 182 |
|
| 183 |
// Otherwise report the first failing check in order. |
| 184 |
foreach ( [ $pretty, $rest, $discovery, $challenge, $user_agent ] as $check ) { |
| 185 |
if ( null === $check ) { |
| 186 |
continue; |
| 187 |
} |
| 188 |
if ( 'ok' !== $check['stage'] ) { |
| 189 |
$result['stage'] = $check['stage']; |
| 190 |
$result['message'] = $check['detail']; |
| 191 |
return $result; |
| 192 |
} |
| 193 |
} |
| 194 |
|
| 195 |
$result['ok'] = true; |
| 196 |
$result['stage'] = 'ok'; |
| 197 |
$result['message'] = sprintf( |
| 198 |
/* translators: %d: number of MCP tools returned. */ |
| 199 |
_n( |
| 200 |
'Connection healthy: the endpoint authenticated, offered OAuth, and returned %d tool.', |
| 201 |
'Connection healthy: the endpoint authenticated, offered OAuth, and returned %d tools.', |
| 202 |
(int) $result['tools_count'], |
| 203 |
'thinkrank' |
| 204 |
), |
| 205 |
(int) $result['tools_count'] |
| 206 |
); |
| 207 |
return $result; |
| 208 |
} |
| 209 |
|
| 210 |
// -- Probes ------------------------------------------------------------ |
| 211 |
|
| 212 |
/** |
| 213 |
* One authenticated JSON-RPC `tools/list` round trip. |
| 214 |
* |
| 215 |
* @param string $url Endpoint to call. |
| 216 |
* @param string $token Connection token. |
| 217 |
* @return array{stage:string,status:?int,tools:?int,authenticated:bool,detail:string} |
| 218 |
*/ |
| 219 |
private static function probe_jsonrpc( string $url, string $token ): array { |
| 220 |
$response = wp_remote_post( |
| 221 |
$url, |
| 222 |
[ |
| 223 |
'timeout' => 10, |
| 224 |
// Don't follow redirects: a 301/302 here IS the finding (the |
| 225 |
// classic http<->https scheme bounce), so surface it verbatim. |
| 226 |
'redirection' => 0, |
| 227 |
'headers' => [ |
| 228 |
'Authorization' => 'Bearer ' . $token, |
| 229 |
'Content-Type' => 'application/json', |
| 230 |
'Accept' => 'application/json', |
| 231 |
], |
| 232 |
'body' => wp_json_encode( |
| 233 |
[ |
| 234 |
'jsonrpc' => '2.0', |
| 235 |
'id' => 1, |
| 236 |
'method' => 'tools/list', |
| 237 |
] |
| 238 |
), |
| 239 |
] |
| 240 |
); |
| 241 |
|
| 242 |
$out = [ |
| 243 |
'stage' => 'ok', |
| 244 |
'status' => null, |
| 245 |
'tools' => null, |
| 246 |
'authenticated' => false, |
| 247 |
'detail' => '', |
| 248 |
]; |
| 249 |
|
| 250 |
if ( is_wp_error( $response ) ) { |
| 251 |
$err = $response->get_error_message(); |
| 252 |
$is_tls = false !== stripos( $err, 'ssl' ) || false !== stripos( $err, 'certificate' ); |
| 253 |
$out['stage'] = $is_tls ? 'tls' : 'unreachable'; |
| 254 |
$out['detail'] = $is_tls |
| 255 |
/* translators: 1: endpoint URL, 2: underlying transport error. */ |
| 256 |
? sprintf( __( '%1$s could not be reached over HTTPS: %2$s. On local/dev sites this is usually a self-signed certificate the AI client must be told to trust.', 'thinkrank' ), $url, $err ) |
| 257 |
/* translators: 1: endpoint URL, 2: underlying transport error. */ |
| 258 |
: sprintf( __( '%1$s could not be reached: %2$s.', 'thinkrank' ), $url, $err ); |
| 259 |
return $out; |
| 260 |
} |
| 261 |
|
| 262 |
$status = (int) wp_remote_retrieve_response_code( $response ); |
| 263 |
$out['status'] = $status; |
| 264 |
|
| 265 |
if ( in_array( $status, [ 301, 302, 307, 308 ], true ) ) { |
| 266 |
$location = (string) wp_remote_retrieve_header( $response, 'location' ); |
| 267 |
$out['stage'] = 'redirect'; |
| 268 |
$out['detail'] = $location |
| 269 |
/* translators: 1: endpoint URL, 2: redirect target URL. */ |
| 270 |
? sprintf( __( '%1$s redirected to %2$s instead of answering. A redirect between HTTP and HTTPS usually means the site address and WordPress address schemes disagree.', 'thinkrank' ), $url, $location ) |
| 271 |
/* translators: %s: endpoint URL. */ |
| 272 |
: sprintf( __( '%s redirected instead of answering, which usually means the site address and WordPress address schemes disagree.', 'thinkrank' ), $url ); |
| 273 |
return $out; |
| 274 |
} |
| 275 |
|
| 276 |
if ( 404 === $status ) { |
| 277 |
$out['stage'] = 'rewrite'; |
| 278 |
$out['detail'] = sprintf( |
| 279 |
/* translators: %s: endpoint URL. */ |
| 280 |
__( '%s returned 404 — WordPress does not know this URL. Re-save Settings → Permalinks to rebuild the rewrite rules.', 'thinkrank' ), |
| 281 |
$url |
| 282 |
); |
| 283 |
return $out; |
| 284 |
} |
| 285 |
|
| 286 |
if ( 401 === $status || 403 === $status ) { |
| 287 |
$out['stage'] = 'auth'; |
| 288 |
$out['detail'] = sprintf( |
| 289 |
/* translators: %s: endpoint URL. */ |
| 290 |
__( '%s rejected the connection token (authentication failed). Rotate the token and reconnect your AI client.', 'thinkrank' ), |
| 291 |
$url |
| 292 |
); |
| 293 |
return $out; |
| 294 |
} |
| 295 |
|
| 296 |
if ( 429 === $status ) { |
| 297 |
$out['stage'] = 'auth'; |
| 298 |
$out['detail'] = sprintf( |
| 299 |
/* translators: %s: endpoint URL. */ |
| 300 |
__( '%s is rate-limiting this server after repeated failed tokens. Wait for the lockout to lapse, then rotate the token and reconnect.', 'thinkrank' ), |
| 301 |
$url |
| 302 |
); |
| 303 |
return $out; |
| 304 |
} |
| 305 |
|
| 306 |
$out['authenticated'] = true; |
| 307 |
$body = json_decode( (string) wp_remote_retrieve_body( $response ), true ); |
| 308 |
$tools = ( is_array( $body ) && isset( $body['result']['tools'] ) && is_array( $body['result']['tools'] ) ) |
| 309 |
? $body['result']['tools'] |
| 310 |
: null; |
| 311 |
|
| 312 |
// An EMPTY tools array counts as a failure, not a pass: that is exactly |
| 313 |
// the shape of #241 — a connection an AI client reports as healthy |
| 314 |
// while it has nothing to call. The abilities snapshot goes into the |
| 315 |
// detail so support can tell "no ThinkRank abilities registered" |
| 316 |
// (foreign Abilities API copy owns the registry) from "the runtime is |
| 317 |
// missing entirely". |
| 318 |
if ( 200 !== $status || null === $tools || [] === $tools ) { |
| 319 |
$out['stage'] = 'no_tools'; |
| 320 |
$out['tools'] = is_array( $tools ) ? count( $tools ) : 0; |
| 321 |
$out['detail'] = sprintf( |
| 322 |
/* translators: 1: endpoint URL, 2: abilities-registry diagnostic summary. */ |
| 323 |
__( '%1$s answered but returned no tool catalog. Confirm the MCP runtime is built and abilities are registered. Diagnostics — %2$s', 'thinkrank' ), |
| 324 |
$url, |
| 325 |
Abilities_Registrar::summary() |
| 326 |
); |
| 327 |
return $out; |
| 328 |
} |
| 329 |
|
| 330 |
$out['tools'] = count( $tools ); |
| 331 |
$out['detail'] = sprintf( |
| 332 |
/* translators: 1: endpoint URL, 2: number of tools. */ |
| 333 |
__( '%1$s authenticated and returned %2$d tools.', 'thinkrank' ), |
| 334 |
$url, |
| 335 |
count( $tools ) |
| 336 |
); |
| 337 |
return $out; |
| 338 |
} |
| 339 |
|
| 340 |
/** |
| 341 |
* Fetch both OAuth discovery documents and confirm they are served and |
| 342 |
* well-formed. An OAuth client reads these before it holds any credential, |
| 343 |
* so a 404 here is invisible to every other check. |
| 344 |
* |
| 345 |
* @return array{stage:string,detail:string} |
| 346 |
*/ |
| 347 |
private static function probe_discovery(): array { |
| 348 |
$documents = []; |
| 349 |
|
| 350 |
// --- The documents clients are POINTED at (must work) ------------- |
| 351 |
// The 401 challenge advertises the REST-served resource metadata, and |
| 352 |
// spec-compliant clients derive the OIDC-suffix form of the AS |
| 353 |
// metadata from our path-based issuer. Neither lives under the site |
| 354 |
// root's /.well-known/ directory, so both survive hosts that |
| 355 |
// intercept that directory at the proxy edge (SiteGround). Each must |
| 356 |
// carry an identifier EXACTLY equal to the one we compute locally — a |
| 357 |
// mere "the key exists" check passes on another plugin's metadata, |
| 358 |
// which is the hijack case the rewrite rules already warn about. |
| 359 |
$primary = [ |
| 360 |
Mcp_OAuth::resource_metadata_url() => [ |
| 361 |
'key' => 'resource', |
| 362 |
'expected' => Mcp_Pairing::site_endpoint(), |
| 363 |
], |
| 364 |
rest_url( 'thinkrank/v1/mcp/oauth/authorization-server' ) => [ |
| 365 |
'key' => 'issuer', |
| 366 |
'expected' => Mcp_OAuth::issuer(), |
| 367 |
], |
| 368 |
Mcp_OAuth::issuer() . '/.well-known/openid-configuration' => [ |
| 369 |
'key' => 'issuer', |
| 370 |
'expected' => Mcp_OAuth::issuer(), |
| 371 |
], |
| 372 |
]; |
| 373 |
|
| 374 |
foreach ( $primary as $url => $spec ) { |
| 375 |
$issue = self::probe_document( $url, $spec, $documents ); |
| 376 |
if ( null !== $issue ) { |
| 377 |
return $issue; |
| 378 |
} |
| 379 |
} |
| 380 |
|
| 381 |
// --- The spec-derived /.well-known/ forms (should work) ----------- |
| 382 |
// A client that ignores the challenge pointer derives these itself |
| 383 |
// (RFC 9728 / RFC 8414 path-insert). Some hosts resolve the root |
| 384 |
// /.well-known/ directory at their proxy as physical files, 404ing |
| 385 |
// before WordPress runs — measurably different from broken rewrites, |
| 386 |
// and fixable by publishing the documents AS physical files. |
| 387 |
$derived = [ |
| 388 |
home_url( '/.well-known/oauth-protected-resource/' . Mcp_Pairing::SITE_ENDPOINT_PATH ) => [ |
| 389 |
'key' => 'resource', |
| 390 |
'expected' => Mcp_Pairing::site_endpoint(), |
| 391 |
], |
| 392 |
home_url( '/.well-known/oauth-authorization-server/' . Mcp_Pairing::SITE_ENDPOINT_PATH ) => [ |
| 393 |
'key' => 'issuer', |
| 394 |
'expected' => Mcp_OAuth::issuer(), |
| 395 |
], |
| 396 |
]; |
| 397 |
|
| 398 |
$root_issue = null; |
| 399 |
foreach ( $derived as $url => $spec ) { |
| 400 |
$root_issue = self::probe_document( $url, $spec, $documents ); |
| 401 |
if ( null !== $root_issue ) { |
| 402 |
break; |
| 403 |
} |
| 404 |
} |
| 405 |
|
| 406 |
if ( null !== $root_issue ) { |
| 407 |
// A document that answers with SOMEONE ELSE'S identity is a plugin |
| 408 |
// conflict poisoning derive-only clients — that stays a hard fail. |
| 409 |
// Only the intercepted/unreachable shapes are softened below. |
| 410 |
if ( 'mismatch' === ( $root_issue['kind'] ?? '' ) ) { |
| 411 |
return $root_issue; |
| 412 |
} |
| 413 |
|
| 414 |
// The host's own trick becomes the fix: if the proxy insists on |
| 415 |
// serving /.well-known/ as physical files, give it physical files. |
| 416 |
/** |
| 417 |
* Filter whether the self-test may publish static /.well-known/ |
| 418 |
* discovery files when the dynamic route is unreachable. |
| 419 |
* |
| 420 |
* @since 1.32.0 |
| 421 |
* |
| 422 |
* @param bool $allowed Defaults to whether the install can host them. |
| 423 |
*/ |
| 424 |
$may_publish = apply_filters( 'thinkrank_mcp_static_discovery_publish', Mcp_Static_Discovery::applicable() ); |
| 425 |
|
| 426 |
$healed = false; |
| 427 |
if ( $may_publish && Mcp_Static_Discovery::publish() ) { |
| 428 |
$healed = true; |
| 429 |
foreach ( $derived as $url => $spec ) { |
| 430 |
if ( null !== self::probe_document( $url, $spec, $documents ) ) { |
| 431 |
$healed = false; |
| 432 |
break; |
| 433 |
} |
| 434 |
} |
| 435 |
} |
| 436 |
|
| 437 |
if ( ! $healed ) { |
| 438 |
// Not fatal on its own any more: the challenge points clients |
| 439 |
// at the REST document (verified above), so the flow survives. |
| 440 |
// Say what is degraded instead of failing the whole check. |
| 441 |
$scheme_issue = self::probe_scheme(); |
| 442 |
if ( null !== $scheme_issue ) { |
| 443 |
$scheme_issue['documents'] = $documents; |
| 444 |
return $scheme_issue; |
| 445 |
} |
| 446 |
return [ |
| 447 |
'stage' => 'ok', |
| 448 |
'documents' => $documents, |
| 449 |
'detail' => __( 'The primary discovery documents are served and correct, but the host intercepts the site root\'s /.well-known/ directory before WordPress runs (common on SiteGround shared hosting), and static files could not be published there. Clients that follow the challenge — ChatGPT, Claude — still connect; a client that only derives the root /.well-known/ URL itself may not. If write access to the site root is possible, granting it lets ThinkRank publish static discovery files that fix this completely.', 'thinkrank' ), |
| 450 |
]; |
| 451 |
} |
| 452 |
} |
| 453 |
|
| 454 |
// Documents agree with us — but they agree on whatever home_url() |
| 455 |
// says, so a site whose stored URL is http:// while it actually serves |
| 456 |
// https:// is self-consistently wrong. Clients connect over https and |
| 457 |
// then reject the http identifier. |
| 458 |
$scheme_issue = self::probe_scheme(); |
| 459 |
if ( null !== $scheme_issue ) { |
| 460 |
$scheme_issue['documents'] = $documents; |
| 461 |
return $scheme_issue; |
| 462 |
} |
| 463 |
|
| 464 |
return [ |
| 465 |
'stage' => 'ok', |
| 466 |
'documents' => $documents, |
| 467 |
'detail' => __( 'All OAuth discovery documents are served and advertise this site\'s MCP endpoint exactly.', 'thinkrank' ), |
| 468 |
]; |
| 469 |
} |
| 470 |
|
| 471 |
/** |
| 472 |
* Fetch and validate one discovery document. Appends what was actually |
| 473 |
* served to $documents either way, so support can read the site's real |
| 474 |
* responses instead of asking the customer for screenshots. |
| 475 |
* |
| 476 |
* @param string $url Document URL. |
| 477 |
* @param array{key:string,expected:string} $spec Identity field + required value. |
| 478 |
* @param array<string,mixed> $documents Accumulator (by reference). |
| 479 |
* @return array{stage:string,documents:array<string,mixed>,detail:string}|null Null when the document is valid. |
| 480 |
*/ |
| 481 |
private static function probe_document( string $url, array $spec, array &$documents ): ?array { |
| 482 |
$response = wp_remote_get( |
| 483 |
$url, |
| 484 |
[ |
| 485 |
'timeout' => 10, |
| 486 |
'redirection' => 0, |
| 487 |
] |
| 488 |
); |
| 489 |
if ( is_wp_error( $response ) ) { |
| 490 |
return [ |
| 491 |
'stage' => 'discovery', |
| 492 |
'kind' => 'unreachable', |
| 493 |
'documents' => $documents, |
| 494 |
'detail' => sprintf( |
| 495 |
/* translators: 1: discovery document URL, 2: transport error. */ |
| 496 |
__( 'The OAuth discovery document %1$s could not be fetched: %2$s. Clients that connect by URL alone cannot authenticate without it.', 'thinkrank' ), |
| 497 |
$url, |
| 498 |
$response->get_error_message() |
| 499 |
), |
| 500 |
]; |
| 501 |
} |
| 502 |
$status = (int) wp_remote_retrieve_response_code( $response ); |
| 503 |
$raw = (string) wp_remote_retrieve_body( $response ); |
| 504 |
$body = json_decode( $raw, true ); |
| 505 |
$documents[ $url ] = is_array( $body ) ? $body : $raw; |
| 506 |
|
| 507 |
if ( 200 !== $status || ! is_array( $body ) || ! isset( $body[ $spec['key'] ] ) ) { |
| 508 |
return [ |
| 509 |
'stage' => 'discovery', |
| 510 |
'kind' => 'invalid', |
| 511 |
'documents' => $documents, |
| 512 |
'detail' => sprintf( |
| 513 |
/* translators: 1: discovery document URL, 2: HTTP status code. */ |
| 514 |
__( 'The OAuth discovery document %1$s returned %2$d instead of valid metadata. Re-save Settings → Permalinks; if it persists, the host may be intercepting the URL before WordPress runs, or another plugin may be claiming it.', 'thinkrank' ), |
| 515 |
$url, |
| 516 |
$status |
| 517 |
), |
| 518 |
]; |
| 519 |
} |
| 520 |
|
| 521 |
$advertised = (string) $body[ $spec['key'] ]; |
| 522 |
if ( $advertised !== $spec['expected'] ) { |
| 523 |
return [ |
| 524 |
'stage' => 'discovery', |
| 525 |
'kind' => 'mismatch', |
| 526 |
'documents' => $documents, |
| 527 |
'detail' => sprintf( |
| 528 |
/* translators: 1: metadata field name, 2: value found in the document, 3: value it should be, 4: discovery document URL. */ |
| 529 |
__( 'The discovery document %4$s advertises %1$s "%2$s" but this site\'s MCP endpoint is "%3$s". RFC 9728 requires an exact match, so clients reject the metadata and report that the server does not implement OAuth. If the two differ only by scheme, a reverse proxy is terminating TLS without passing X-Forwarded-Proto; otherwise another plugin is serving this URL.', 'thinkrank' ), |
| 530 |
$spec['key'], |
| 531 |
$advertised, |
| 532 |
$spec['expected'], |
| 533 |
$url |
| 534 |
), |
| 535 |
]; |
| 536 |
} |
| 537 |
|
| 538 |
return null; |
| 539 |
} |
| 540 |
|
| 541 |
/** |
| 542 |
* Catch the reverse-proxy scheme trap: WordPress stores an http:// home |
| 543 |
* URL, so every advertised OAuth identifier is http://, while the site is |
| 544 |
* really served over https://. Everything is internally consistent, so no |
| 545 |
* comparison against our own values can see it — the only tell is that the |
| 546 |
* https:// variant of the endpoint answers too. |
| 547 |
* |
| 548 |
* @return array{stage:string,detail:string}|null Null when nothing is wrong. |
| 549 |
*/ |
| 550 |
private static function probe_scheme(): ?array { |
| 551 |
$endpoint = Mcp_Pairing::site_endpoint(); |
| 552 |
if ( 'https' === wp_parse_url( $endpoint, PHP_URL_SCHEME ) ) { |
| 553 |
return null; |
| 554 |
} |
| 555 |
|
| 556 |
$secure = set_url_scheme( $endpoint, 'https' ); |
| 557 |
if ( null === self::probe_status( $secure, null ) ) { |
| 558 |
// No HTTPS at all. A plain-HTTP site is its own (reported) problem, |
| 559 |
// not the proxy misconfiguration this check is for. |
| 560 |
return null; |
| 561 |
} |
| 562 |
|
| 563 |
return [ |
| 564 |
'stage' => 'discovery', |
| 565 |
'detail' => sprintf( |
| 566 |
/* translators: 1: http endpoint URL advertised, 2: https endpoint URL that also answers. */ |
| 567 |
__( 'The discovery documents advertise %1$s, but %2$s answers as well — WordPress is storing an http:// site address behind a proxy that terminates TLS. AI clients connect over https and reject the http identifier as a mismatch. Fix the Site Address in Settings → General, or have the proxy send X-Forwarded-Proto.', 'thinkrank' ), |
| 568 |
$endpoint, |
| 569 |
$secure |
| 570 |
), |
| 571 |
]; |
| 572 |
} |
| 573 |
|
| 574 |
/** |
| 575 |
* Confirm an unauthenticated call answers 401 WITH the RFC 9728 |
| 576 |
* WWW-Authenticate challenge. A client that connects by URL alone has |
| 577 |
* nothing else to discover OAuth from — a bare 401, or any other status, |
| 578 |
* reads to it as "this server does not implement OAuth". |
| 579 |
* |
| 580 |
* @param string $endpoint Pretty endpoint URL. |
| 581 |
* @param string $fallback REST fallback URL. |
| 582 |
* @return array{stage:string,detail:string} |
| 583 |
*/ |
| 584 |
private static function probe_challenge( string $endpoint, string $fallback ): array { |
| 585 |
$answered = false; |
| 586 |
|
| 587 |
foreach ( [ $endpoint, $fallback ] as $url ) { |
| 588 |
$response = wp_remote_post( |
| 589 |
$url, |
| 590 |
[ |
| 591 |
'timeout' => 10, |
| 592 |
'redirection' => 0, |
| 593 |
'headers' => [ |
| 594 |
'Content-Type' => 'application/json', |
| 595 |
'Accept' => 'application/json', |
| 596 |
], |
| 597 |
'body' => wp_json_encode( |
| 598 |
[ |
| 599 |
'jsonrpc' => '2.0', |
| 600 |
'id' => 1, |
| 601 |
'method' => 'initialize', |
| 602 |
'params' => [], |
| 603 |
] |
| 604 |
), |
| 605 |
] |
| 606 |
); |
| 607 |
if ( is_wp_error( $response ) ) { |
| 608 |
continue; // Reachability is the other checks' job. |
| 609 |
} |
| 610 |
$answered = true; |
| 611 |
|
| 612 |
$status = (int) wp_remote_retrieve_response_code( $response ); |
| 613 |
$challenge = (string) wp_remote_retrieve_header( $response, 'www-authenticate' ); |
| 614 |
|
| 615 |
if ( 401 !== $status ) { |
| 616 |
return [ |
| 617 |
'stage' => 'challenge', |
| 618 |
'detail' => sprintf( |
| 619 |
/* translators: 1: endpoint URL, 2: HTTP status code. */ |
| 620 |
__( 'An unauthenticated call to %1$s answered %2$d instead of 401. Clients that connect by URL alone need the 401 challenge to start the OAuth flow.', 'thinkrank' ), |
| 621 |
$url, |
| 622 |
$status |
| 623 |
), |
| 624 |
]; |
| 625 |
} |
| 626 |
if ( '' === $challenge ) { |
| 627 |
return [ |
| 628 |
'stage' => 'challenge', |
| 629 |
'detail' => sprintf( |
| 630 |
/* translators: %s: endpoint URL. */ |
| 631 |
__( '%s answered 401 but sent no WWW-Authenticate header — a security plugin or proxy is likely stripping it. Clients that connect by URL alone will report that this server does not implement OAuth.', 'thinkrank' ), |
| 632 |
$url |
| 633 |
), |
| 634 |
]; |
| 635 |
} |
| 636 |
|
| 637 |
// The challenge is only useful if the URL inside it resolves — |
| 638 |
// that URL is the client's entire entry point into the flow. |
| 639 |
if ( ! preg_match( '/resource_metadata="([^"]+)"/i', $challenge, $m ) ) { |
| 640 |
return [ |
| 641 |
'stage' => 'challenge', |
| 642 |
'detail' => sprintf( |
| 643 |
/* translators: 1: endpoint URL, 2: the WWW-Authenticate header value received. */ |
| 644 |
__( '%1$s sent a WWW-Authenticate header with no resource_metadata URL (%2$s). Clients have nowhere to look up this site\'s OAuth metadata.', 'thinkrank' ), |
| 645 |
$url, |
| 646 |
$challenge |
| 647 |
), |
| 648 |
]; |
| 649 |
} |
| 650 |
|
| 651 |
$metadata_url = $m[1]; |
| 652 |
$metadata = wp_remote_get( |
| 653 |
$metadata_url, |
| 654 |
[ |
| 655 |
'timeout' => 10, |
| 656 |
'redirection' => 2, // A host-level redirect to the real doc is fine. |
| 657 |
] |
| 658 |
); |
| 659 |
$reachable = ! is_wp_error( $metadata ) |
| 660 |
&& 200 === (int) wp_remote_retrieve_response_code( $metadata ) |
| 661 |
&& is_array( json_decode( (string) wp_remote_retrieve_body( $metadata ), true ) ); |
| 662 |
|
| 663 |
if ( ! $reachable ) { |
| 664 |
return [ |
| 665 |
'stage' => 'challenge', |
| 666 |
'detail' => sprintf( |
| 667 |
/* translators: 1: resource_metadata URL from the challenge header, 2: endpoint URL. */ |
| 668 |
__( 'The challenge from %2$s points at %1$s, but that URL does not return OAuth metadata. This is the first thing a client fetches, so the connection fails there. A security plugin or edge rule blocking the REST API for visitors is the usual cause.', 'thinkrank' ), |
| 669 |
$metadata_url, |
| 670 |
$url |
| 671 |
), |
| 672 |
]; |
| 673 |
} |
| 674 |
} |
| 675 |
|
| 676 |
// Neither URL answered at all. Reporting `ok` here would be the exact |
| 677 |
// false pass this test exists to prevent — an unreachable endpoint is |
| 678 |
// not a passing challenge. The other checks name the reachability |
| 679 |
// failure, so this one only has to refuse to claim success. |
| 680 |
if ( ! $answered ) { |
| 681 |
return [ |
| 682 |
'stage' => 'challenge', |
| 683 |
'detail' => __( 'The OAuth challenge could not be checked because the endpoint did not answer. Fix the connection error above and re-run the test.', 'thinkrank' ), |
| 684 |
]; |
| 685 |
} |
| 686 |
|
| 687 |
return [ |
| 688 |
'stage' => 'ok', |
| 689 |
'detail' => __( 'Unauthenticated calls answer with the OAuth challenge, so URL-only clients can authenticate.', 'thinkrank' ), |
| 690 |
]; |
| 691 |
} |
| 692 |
|
| 693 |
/** |
| 694 |
* Detect a host that answers WordPress but refuses AI clients by |
| 695 |
* User-Agent. Replays the unauthenticated probe under the UAs a real MCP |
| 696 |
* backend sends and compares against the baseline; a 403/406/503 that the |
| 697 |
* baseline did not get is a bot filter, not a plugin problem. |
| 698 |
* |
| 699 |
* Blind spot worth stating plainly: this runs from the server's own IP, |
| 700 |
* which host firewalls usually trust, so it catches UA filtering but NOT |
| 701 |
* an IP-range block of the AI vendor. A green result here does not prove |
| 702 |
* an external client can connect. |
| 703 |
* |
| 704 |
* @param string $endpoint Pretty endpoint URL. |
| 705 |
* @return array{stage:string,detail:string}|null Null when it could not run. |
| 706 |
*/ |
| 707 |
private static function probe_user_agent( string $endpoint ): ?array { |
| 708 |
$baseline = self::probe_status( $endpoint, null ); |
| 709 |
if ( null === $baseline ) { |
| 710 |
return null; // Endpoint unreachable — the other checks own that. |
| 711 |
} |
| 712 |
|
| 713 |
foreach ( self::CLIENT_USER_AGENTS as $agent ) { |
| 714 |
$status = self::probe_status( $endpoint, $agent ); |
| 715 |
if ( null === $status || $status === $baseline ) { |
| 716 |
continue; |
| 717 |
} |
| 718 |
// A different status is only damning when it is a refusal. An MCP |
| 719 |
// answer (401 challenge / 200 / 202) under any UA is fine. |
| 720 |
if ( in_array( $status, [ 200, 202, 401 ], true ) ) { |
| 721 |
continue; |
| 722 |
} |
| 723 |
return [ |
| 724 |
'stage' => 'ua_filter', |
| 725 |
'doc_url' => self::HOSTING_DOC_URL, |
| 726 |
'detail' => sprintf( |
| 727 |
/* translators: 1: user agent string, 2: HTTP status returned for it, 3: HTTP status returned for WordPress's own user agent. */ |
| 728 |
__( 'The endpoint answered %3$d for WordPress but %2$d for an AI client\'s User-Agent (%1$s). ThinkRank deliberately tests with the generic agents real MCP backends send; this refusal means a security plugin, firewall or host-level "block bad bots" rule (SiteGround\'s edge protection does this) will also refuse the real AI client. Ask the host to exempt the MCP and /.well-known/ paths, or allowlist these User-Agents.', 'thinkrank' ), |
| 729 |
$agent, |
| 730 |
$status, |
| 731 |
$baseline |
| 732 |
), |
| 733 |
]; |
| 734 |
} |
| 735 |
|
| 736 |
return [ |
| 737 |
'stage' => 'ok', |
| 738 |
'detail' => __( 'The endpoint answers AI-client User-Agents the same way it answers WordPress, so no bot filter is blocking them. This cannot see an IP-level block of the AI vendor.', 'thinkrank' ), |
| 739 |
]; |
| 740 |
} |
| 741 |
|
| 742 |
// -- Helpers ----------------------------------------------------------- |
| 743 |
|
| 744 |
/** |
| 745 |
* Status code of one unauthenticated probe, or null if it never answered. |
| 746 |
* |
| 747 |
* @param string $url Endpoint to call. |
| 748 |
* @param string|null $agent User-Agent to send, or null for WordPress's own. |
| 749 |
* @return int|null |
| 750 |
*/ |
| 751 |
private static function probe_status( string $url, ?string $agent ): ?int { |
| 752 |
$args = [ |
| 753 |
'timeout' => 10, |
| 754 |
'redirection' => 0, |
| 755 |
'headers' => [ |
| 756 |
'Content-Type' => 'application/json', |
| 757 |
'Accept' => 'application/json', |
| 758 |
], |
| 759 |
'body' => wp_json_encode( |
| 760 |
[ |
| 761 |
'jsonrpc' => '2.0', |
| 762 |
'id' => 1, |
| 763 |
'method' => 'initialize', |
| 764 |
'params' => [], |
| 765 |
] |
| 766 |
), |
| 767 |
]; |
| 768 |
if ( null !== $agent ) { |
| 769 |
$args['user-agent'] = $agent; |
| 770 |
} |
| 771 |
|
| 772 |
$response = wp_remote_post( $url, $args ); |
| 773 |
if ( is_wp_error( $response ) ) { |
| 774 |
return null; |
| 775 |
} |
| 776 |
return (int) wp_remote_retrieve_response_code( $response ); |
| 777 |
} |
| 778 |
|
| 779 |
/** |
| 780 |
* Shape one check for the UI list. |
| 781 |
* |
| 782 |
* @param string $id Check id. |
| 783 |
* @param string $label Human label. |
| 784 |
* @param string $stage Resulting stage ('ok' when it passed). |
| 785 |
* @param string $detail Explanatory line. |
| 786 |
* @param string $doc_url Optional docs page for a failure the user has to |
| 787 |
* fix outside WordPress. Omitted when empty. |
| 788 |
* @return array{id:string,label:string,ok:bool,detail:string,doc_url?:string} |
| 789 |
*/ |
| 790 |
private static function check( string $id, string $label, string $stage, string $detail, string $doc_url = '' ): array { |
| 791 |
$check = [ |
| 792 |
'id' => $id, |
| 793 |
'label' => $label, |
| 794 |
'ok' => 'ok' === $stage, |
| 795 |
'detail' => $detail, |
| 796 |
]; |
| 797 |
|
| 798 |
if ( '' !== $doc_url ) { |
| 799 |
$check['doc_url'] = $doc_url; |
| 800 |
} |
| 801 |
|
| 802 |
return $check; |
| 803 |
} |
| 804 |
} |
| 805 |
|
