AI CyberGuard - Firewall, Malware Scanner & Login Security
| 开发者 | yosikrivo |
|---|---|
| 更新时间 | 2026年7月29日 17:41 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.0 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
详情介绍:
AI CyberGuard is one complete, AI-powered security solution for WordPress — combining a firewall, malware scanner, login protection, file-integrity monitoring and safety snapshots in a single plugin, with clear, actionable guidance instead of cryptic alerts.
Key Features
- Firewall (Lite) - front-end URL & query-string filter - a front-end request filter (Monitor or Enforce) that acts on your local IP blocklist plus an optional small set of high-confidence URL/query signatures (SQL injection, path traversal, local file inclusion, XSS, PHP code injection and reconnaissance probes). It inspects only the request URL and query string on front-end page loads. It does NOT inspect POST bodies, request headers or cookies, and it does not filter the admin area, AJAX, XML-RPC, WP-Cron or the login entry point; logged-in admins and whitelisted IPs are always exempt. REST API requests ARE inspected, by the same URL and query-string checks as any front-end request: WordPress does not mark a request as REST until after the firewall has run, so it cannot be exempted. With the signature rules off (the default) that means your IP blocklist applies to /wp-json/ too. Test Enforce mode on staging. (PRO adds an always-on, auto-updating WAF that inspects POST bodies and headers in real time.)
- Login Security - Brute-force lockout, privacy-first CAPTCHA (honeypot / Cloudflare Turnstile), and breached-password checks
- Two-Factor Authentication (2FA) - Per-user TOTP (Google Authenticator / Authy) with backup codes
- Safety Snapshots - Point-in-time copies of critical files; to restore, download a verified clean copy and place it back yourself (automatic in-place restore is disabled for security)
- Hardening & Hide Login - One-click hardening (XML-RPC, user enumeration, file editor) plus a custom login URL
- Security Hub & Score - Security grade, Quick Wins, 7-day event log, weekly email reports, and an on-device AI Assistant
- File Integrity Monitoring - Create a baseline of your files and detect any changes
- Scheduled Scans with Malware Alerts - Automatic background scans compare your files against the baseline and then malware-scan the files that changed or were added, so a backdoor that arrives on your site is found without you having to remember to scan. Findings appear on the dashboard; if you switch Email Alerts on (they are off by default) a confirmed threat is also emailed to you straight away, without waiting for the usual rate limit. Bounds, so you know what a quiet scan does and does not mean: a scheduled run inspects up to 300 changed files per run (the rest are reported as not-yet-inspected and picked up on later runs) and skips files over 2 MB - both are stated in the alert itself. To check every file on the site, run a Malware Scan from the dashboard.
- Malware Pattern Detection - Scan for known malicious code patterns: 100+ built-in patterns, expanded to thousands via the optional weekly threat feed
- Repository Comparison - Compare plugins against WordPress.org checksums
- File Change Details - See which files changed, when, and their before/after checksums
- Clear Reports - Human-friendly explanations, not cryptic alerts
- Smart Filtering - Reduces false positives with context-aware analysis
- Database Scanner - Check for suspicious content in your database
- Security Checklist - Quick overview of your site's security status
- MalwareBazaar Integration - Verify threats against known malware database
- Create Baseline - AI CyberGuard takes a snapshot of your current files
- Run Scans - Use Quick Malware Scan for a focused malware check, Deep Scan for the entire site, and Regular or Full Scan for file changes
- Review Changes - See exactly what changed, when, and why it matters
- Take Action - Clear guidance on what to do with each finding
- Triggered: lazily, the first time you perform a user action that needs the GuardPress server - running an AI Analysis (after enabling AI), or enabling / refreshing the Threat Feed. Simply switching AI on in Settings does not register on its own; the first Analyze does.
- Data sent: random installation UUID + plugin version
- Data NOT sent: site URL, file contents, personal data, email
- Purpose: issues a per-installation HMAC secret used for all later API authentication (so the plugin is not bundled with a shared salt)
- Privacy: https://guardpress.ai/privacy
- Status: Must be enabled manually in Settings → Privacy & External Requests
- Purpose: Download latest malware signatures
- Data sent: HMAC-signed request carrying installation UUID + plugin version
- Data NOT sent: site URL, file contents, personal data
- Status: Must be enabled manually in Settings → Privacy & External Requests
- Purpose: Deep code analysis for complex threats
- Data sent, when you click "Analyze": the file's content (sensitive credentials redacted before transmission), plus the file name, its path relative to your WordPress root, the file type/extension, and your WordPress version. NOT sent: the absolute server path, host name, or any OS username.
- Requires explicit user action - never automatic
- Purpose: Verify WordPress core and installed-plugin file integrity (checksums + ZIP comparison)
- Data sent: WordPress core version, plugin slugs and versions
- These are the official WordPress.org APIs
- Opt-in: runs only during a malware scan, and only while the Threat Intelligence Feed is enabled (off by default)
- Purpose: Confirm known malware by hash lookup
- Data sent: SHA256 hash of suspicious files only
- Data NOT sent: file contents, filenames, site URL
- Privacy: https://abuse.ch/privacy
- This plugin does NOT contact URLhaus and does NOT send any URL from your files anywhere. Earlier versions of this document described a URLhaus lookup; that lookup is not performed by the plugin, so the entry has been removed rather than left as an inaccurate disclosure.
- Purpose: Look up the country of blocked IPs so the dashboard "Live Attack Map" can plot them (display only)
- Opt-in: OFF by default. No IP is ever sent unless you enable "Attack-map location lookup (GeoIP)" under Settings & Security → Privacy & External Requests. While off, the map shows only countries already cached locally.
- Data sent: IP addresses that have already been blocked by your site (not arbitrary IPs), only when you open the dashboard, rate-limited
- Data NOT sent: file contents, account data, site URL
- Privacy: https://ipinfo.io/privacy
- Status: Used only if you enable the breached-password check
- Purpose: Warn when a login password appears in a known data breach
- Data sent: Only the first 5 characters of the password's SHA-1 hash (k-anonymity model) - never the password, username, or full hash
- Data NOT sent: the password, the full hash, site URL, personal data
- Privacy: https://haveibeenpwned.com/Privacy
- Status: Used only if you enable Turnstile CAPTCHA on the login form
- Purpose: Privacy-friendly bot / CAPTCHA verification on login
- In the visitor's browser: Cloudflare's Turnstile script and challenge widget load from challenges.cloudflare.com using your public site key, the same as any site that embeds Turnstile
- Sent from your server (only to verify a solved challenge): your private secret key and the one-time challenge-response token
- Data NOT sent from your server: the visitor's IP address, file contents, passwords, usernames, or other personal data
- Privacy: https://www.cloudflare.com/privacypolicy/
WP_HTTP_BLOCK_EXTERNAL constant for sites that prefer to disable all outbound traffic. The one exception is Cloudflare Turnstile (item 9): its script and challenge load in the visitor's browser (not via the WordPress HTTP API), only when you enable Turnstile, and are therefore not governed by WP_HTTP_BLOCK_EXTERNAL.
安装:
- Upload the
ai-cyberguardfolder to/wp-content/plugins/ - Activate the plugin through the 'Plugins' menu in WordPress
- Go to AI CyberGuard in your admin menu
- Run a Quick Malware Scan
- Review the scan results and confirm that the site is clean
- Click Create Baseline to record the current clean state
屏幕截图:
常见问题:
Will this slow down my site?
Scans run only when you request one or on the scheduled cron job, never on a visitor page load. The firewall does run on front-end requests - that is what a firewall is - but it only inspects the request URL and query string, with no database scan and no file scan. If you would rather it did nothing at all on the front end, set it to Monitor mode or turn it off in Settings.
What is the difference between the scan types?
AI CyberGuard includes two different groups of scans:
- Quick Malware Scan: Performs a focused malware check of the most relevant executable WordPress areas.
- Deep Scan (entire site): Performs a broader malware scan across the full WordPress installation and may take longer.
- Regular Scan: Compares the usual WordPress files against your saved file-integrity baseline.
- Full Scan: Performs the broader file-change comparison across all files and is slower.
How accurate is the malware detection?
AI CyberGuard aims for high-confidence detection with documented boundaries rather than a promise to catch everything - no scanner can honestly make that promise. Concretely: when malicious code hides the name of the function it calls behind a string transformation, the scanner resolves a curated list of those transformations (decoders, case and trim helpers, string and path builders, and array literals) and reports the result as a confirmed threat. Transformations outside that list, and code whose behaviour depends on values computed earlier at runtime, are documented boundaries rather than covered cases. Findings are separated into confirmed threats (a known-malware hash, or code that provably executes attacker-controlled input) and review items (patterns that are worth a look but are also used by legitimate plugins, such as a code-snippet plugin that evaluates code by design). Only confirmed threats are counted as malware, so a site full of legitimate premium plugins does not read as a wall of alarms. Files that could not be read are reported as a coverage gap rather than silently counted clean.
Does AI CyberGuard automatically delete files?
No. AI CyberGuard is a detection tool - it shows you what changed and why it might be suspicious. You decide what action to take. This prevents accidental deletion of legitimate files.
Can I compare my theme files against a repository?
Theme checksum comparison is not available because most themes are not hosted on WordPress.org. Themes are scanned using pattern-based detection instead.
What if I get false positives?
Use the "Ignore" feature to exclude known legitimate files. You can also configure exclusion patterns for directories like caches and logs.
更新日志:
1.0.2
- Performance: the security engine now loads only where it is needed, so front-end pages are lighter and use less memory.
- Minor improvements.