agentleFS
wpultimatesecuritywpultimatesecurity/WordPress-Security-Skills/skills/ajax-security/SKILL.md

ajax-security

Skill45 starsChanged 54 days agoMITView on GitHub

Use when registering or handling WordPress AJAX over admin-ajax.php - wp_ajax_{action} / wp_ajax_nopriv_{action} hooks, JavaScript that posts to admin_url('admin-ajax.php'), or wp.apiFetch / fetch calls to custom actions. Verifies the nonce with check_ajax_referer, gates the action with current_user_can, unslashes and sanitizes every field, and replies with wp_send_json_success / wp_send_json_error. Prevents CSRF, broken access control, and injection on the AJAX surface.

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this skill whenever code registers or handles an AJAX action through WordPress's admin-ajax.php endpoint:

    line 21
  2. Very specific:

    If the request changes state or exposes privileged data, it needs a nonce and a capability check.

    line 29
  3. Very specific:

    Check current_user_can() immediately after the nonce.

    line 65

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

AJAX security

When to use this skill

Use this skill whenever code registers or handles an AJAX action through WordPress's admin-ajax.php endpoint:

  • add_action( 'wp_ajax_{action}', ... ) for logged-in users.
  • add_action( 'wp_ajax_nopriv_{action}', ... ) for logged-out visitors.
  • JavaScript posting to admin_url( 'admin-ajax.php' ) with jQuery, fetch, or wp.apiFetch.
  • Returning JSON, HTML fragments, or any server-computed data from an AJAX handler.

If the request changes state or exposes privileged data, it needs a nonce and a capability check. wp_ajax_nopriv_* is intentionally unauthenticated — never use it for admin actions.

Related: see the nonces-csrf-protection skill for the full nonce lifecycle and the rest-api-security skill for REST endpoints.

Core principles (and why they matter)

  1. wp_ajax_* vs wp_ajax_nopriv_* are different trust boundaries. nopriv fires for anonymous visitors; use it only for truly public actions. Privileged actions must use wp_ajax_*.
  2. Nonce + capability for privileged actions. check_ajax_referer() checks a CSRF token, not authenticated origin or authorization; current_user_can() checks permission.
  3. Sanitize every input field. $_POST values in AJAX are just as attacker-controllable as any other request. wp_unslash() then sanitize to type before use.
  4. Exit through wp_send_json_* or wp_die(). Never echo raw output and fall through. JSON responses keep the contract explicit and prevent accidental HTML/PHP leakage.
  5. Scope the nonce to the action. A nonce for my_plugin_delete_item_42 cannot be replayed to delete item 99.
  6. Fail closed. Any failed check returns a JSON error or dies; the handler never continues to act.
  7. Throttle anonymous and expensive actions before the work. Public nonces can be obtained and replayed by attackers; they are not rate limits. Transient counters are best-effort load shedding only: read/increment/write is non-atomic and cached entries can disappear early. Strict limits need an atomic shared backend or an edge/server limiter with defined windows, trusted client identity, and failure policy.

Step-by-step implementation

  1. Register the handler with the correct hook (wp_ajax_* for authenticated, wp_ajax_nopriv_* for public).
  2. Enqueue the script and pass the AJAX URL + nonce via wp_localize_script() or wp_add_inline_script().
  3. In the handler, verify the nonce first with check_ajax_referer( $action, $query_arg, false ). Passing false lets you return a clean JSON error instead of dying with -1.
  4. Check current_user_can() immediately after the nonce.
  5. wp_unslash() and sanitize every $_POST / $_GET field, using the right sanitizer for the type (absint, sanitize_text_field, sanitize_key, sanitize_email, etc.).
  6. Enforce the abuse policy before queries, mail, or other expensive actions; return wp_send_json_error( ..., 429 ) when rejected. Use an atomic limiter where bypass would create a security or cost risk; a transient is only a best-effort supplement.
  7. Perform the authorized action.
  8. Return wp_send_json_success() or wp_send_json_error() and stop.

Supporting references

ReferenceLoad when
AJAX security checklistBefore final verification of AJAX security controls.
Secure AJAX handlerImplementing the privileged admin-ajax.php profile-update flow with nonce, capability, input validation, and JSON responses.

Common AI mistakes / anti-patterns

Mistake 1 — No nonce verification

// ❌ Insecure: any site can POST to this action and trigger the handler.
add_action( 'wp_ajax_my_plugin_vote', function () {
    $post_id = absint( $_POST['post_id'] );
    update_post_meta( $post_id, '_votes', get_post_meta( $post_id, '_votes', true ) + 1 );
    wp_send_json_success();
} );
// ✅ Secure: verify nonce, then capability, then act.
add_action( 'wp_ajax_my_plugin_vote', 'my_plugin_ajax_vote' );
function my_plugin_ajax_vote() {
    if ( ! check_ajax_referer( 'my_plugin_vote', 'nonce', false ) ) {
        wp_send_json_error( array( 'message' => __( 'Security check failed.', 'my-plugin' ) ), 403 );
    }
    if ( ! current_user_can( 'edit_posts' ) ) {
        wp_send_json_error( array( 'message' => __( 'Forbidden.', 'my-plugin' ) ), 403 );
    }
    $post_id = isset( $_POST['post_id'] ) ? absint( wp_unslash( $_POST['post_id'] ) ) : 0;
    if ( $post_id <= 0 ) {
        wp_send_json_error( array( 'message' => __( 'Invalid post.', 'my-plugin' ) ), 400 );
    }
    update_post_meta( $post_id, '_votes', get_post_meta( $post_id, '_votes', true ) + 1 );
    wp_send_json_success();
}

Mistake 2 — Using wp_ajax_nopriv_* for a privileged action

// ❌ Insecure: anonymous users can now save settings.
add_action( 'wp_ajax_nopriv_my_plugin_save_settings', 'my_plugin_save_settings' );
// ✅ Secure: privileged actions use wp_ajax_* only.
add_action( 'wp_ajax_my_plugin_save_settings', 'my_plugin_save_settings' );
function my_plugin_save_settings() {
    check_ajax_referer( 'my_plugin_save_settings', 'nonce' );
    if ( ! current_user_can( 'manage_options' ) ) {
        wp_send_json_error( array( 'message' => __( 'Forbidden.', 'my-plugin' ) ), 403 );
    }
    // ...sanitize + save...
    wp_send_json_success();
}

Mistake 3 — Echoing raw $_POST or building HTML without escaping

// ❌ Insecure: reflected XSS.
echo '<div>' . $_POST['message'] . '</div>';
// ✅ Secure: escape at the point of output, even inside a JSON payload.
$message = sanitize_text_field( wp_unslash( $_POST['message'] ?? '' ) );
wp_send_json_success( array( 'html' => '<div>' . esc_html( $message ) . '</div>' ) );

Mistake 4 — Forgetting wp_unslash()

// ❌ Risky: WordPress slashes superglobals; data may contain escaped quotes.
$label = sanitize_text_field( $_POST['label'] );
// ✅ Secure: unslash first, then sanitize.
$label = isset( $_POST['label'] ) ? sanitize_text_field( wp_unslash( $_POST['label'] ) ) : '';

Mistake 5 — Trusting is_user_logged_in() instead of a capability

// ❌ Insecure: every logged-in user, including subscribers, can run this.
if ( is_user_logged_in() ) {
    delete_option( 'my_plugin_config' );
}
// ✅ Secure: check a real capability.
if ( ! current_user_can( 'manage_options' ) ) {
    wp_send_json_error( array( 'message' => __( 'Forbidden.', 'my-plugin' ) ), 403 );
}

Mistake 6 — Unthrottled public action

// ❌ Insecure: anonymous endpoint, no rate limit — unbounded cost per visitor.
add_action( 'wp_ajax_nopriv_my_plugin_search', 'my_plugin_ajax_search' );
function my_plugin_ajax_search() {
    $q = isset( $_POST['q'] ) ? sanitize_text_field( wp_unslash( $_POST['q'] ) ) : '';
    wp_send_json_success( my_plugin_run_expensive_query( $q ) );
}
// Best-effort only: public, read-only search with bounded output.
add_action( 'wp_ajax_nopriv_my_plugin_search', 'my_plugin_ajax_search' );
function my_plugin_ajax_search() {
    if ( ! isset( $_POST['q'] ) || ! is_string( $_POST['q'] ) || strlen( $_POST['q'] ) > 200 ) {
        wp_send_json_error( array( 'message' => 'Invalid query.' ), 400 );
    }
    $q = sanitize_text_field( wp_unslash( $_POST['q'] ) );
    if ( '' === $q ) {
        wp_send_json_error( array( 'message' => 'Query required.' ), 400 );
    }
    // Fixed minute buckets; counters can race or be evicted. Not a strict quota.
    $ip     = $_SERVER['REMOTE_ADDR'] ?? '';
    $window = intdiv( time(), MINUTE_IN_SECONDS );
    $key    = 'my_plugin_rl_' . wp_hash( 'search|' . $ip . '|' . $window );
    $count  = (int) get_transient( $key );
    if ( $count >= 20 ) {
        wp_send_json_error( array( 'message' => __( 'Too many requests.', 'my-plugin' ) ), 429 );
    }
    set_transient( $key, $count + 1, MINUTE_IN_SECONDS );

    // The counter is checked before the query, never after sending the response.
    $ids = get_posts( array(
        's'           => $q,
        'post_type'   => 'post',
        'post_status' => 'publish',
        'numberposts' => 10,
        'fields'      => 'ids',
    ) );
    wp_send_json_success( $ids );
}

This read-only public search does not use a nonce as an abuse control. State-changing or privileged handlers still need their CSRF and capability checks. Fixed windows allow bursts across boundaries; parallel requests can lose counter increments, and cache eviction/storage failure can reset this best-effort counter. Do not use it as the sole control for mail sending, paid APIs, voting integrity, or brute-force defense.

REMOTE_ADDR identifies the direct peer: behind a proxy it may identify the proxy, not the visitor, and shared NATs group users together. Only trust forwarded addresses after a configured trusted proxy strips client-supplied headers and supplies a validated address; never read arbitrary X-Forwarded-For directly. Strict distributed limits need an atomic shared operation (including expiry, e.g. a Redis script), or an edge/server limiter on all relevant routes. See authentication-session-security for login-specific considerations.

Correct code examples

A complete, copy-paste-ready AJAX handler (PHP + JavaScript) is in references/secure-ajax-handler.php.

Checklist

  • The action is registered on wp_ajax_* for privileged flows or wp_ajax_nopriv_* only when the action is truly public.
  • The handler verifies the nonce with check_ajax_referer() before acting.
  • A current_user_can() check runs immediately after nonce verification.
  • Every $_POST / $_GET field is wp_unslash()-ed and sanitized to type.
  • The handler exits via wp_send_json_success(), wp_send_json_error(), or wp_die().
  • Output inside JSON responses is escaped for its context (esc_html, esc_attr, esc_url).
  • The nonce action string is specific and scoped (e.g., includes an object id).
  • Secrets are never returned to the browser or logged.
  • Abuse controls run before expensive actions; strict limits use atomic shared or edge enforcement, not transients.
  • Client identity accounts for trusted proxies/shared IPs; arbitrary forwarded headers are not trusted.

Official references

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.