Website Creation Wizard

Configure and launch a new satellite model directory in minutes.

🔍 About Platform Semantic Search

This tool searches through all ingested database records for this tenant (including models, blog posts, videos, and galleries) using natural language meaning rather than simple keyword matches. Use it to verify indexed content (e.g. "blonde online stripchat") and to retrieve unique UUIDs from the details panel on the right to assign models to blog posts or debug active entity attributes.

📜 Logs

Currently Armed

🧭 Verdict

This section shows an AI's interpretation of the report above (not raw data) - nothing here is generated until you ask for one.

🩺 Server Health

🧱 Nginx Blocking Layer

Requests nginx rejected outright ($should_block) before they ever reached the site - logged to blocked_bots.log.

🔥 PHP Firewall Layer (Apache 429s)

Requests that got PAST nginx but were then rate-limited (HTTP 429) at the application layer, per site.

Attack URL Patterns Detected

✅ Good Bots & Humans Check

For each named good-bot, how it was actually treated by real site traffic (not the blocked_bots log) - response codes it got when it visited pages normally. A healthy pattern is mostly 200s; frequent 429s here means even a legitimate crawler is getting rate-limited.

Note: real Googlebot/Bingbot/etc are usually allowlisted ($is_good_bot_ip) and bypass the PoL cookie challenge entirely - this table is about response codes, not the PoL gate.

📜 History & Trend

Past reports, newest first - nothing here is ever deleted or overwritten. Click a row to view that report above (with its own delta vs the one before it); use "◂ Back to latest" up top to return.

A configurable layer on top of the always-on nginx rules (see the Bot Analytics panel for those) - every rule below starts in monitor and stays there until you explicitly change it. In monitor mode a rule computes and logs its score exactly like it would if enforced, but never touches a real request - the numbers below show what it WOULD have done.

How this works, in plain terms: each rule below checks for one specific pattern (e.g. a WordPress scan, an old browser). When a rule's pattern matches a request, it adds its score to that request's total. The Scoring thresholds rule (near the bottom of the list) then decides what happens based on that total - low enough is served normally, higher gets logged, higher still gets sent to prove they're human, and the highest band gets rejected. A rule only actually enforces (challenge/block) real visitors if you switch that specific rule's mode away from monitor - the threshold total alone changes nothing by itself.

Scoring rules

🚫 Auto-block state (escalating)

Stateful - an IP only appears here once it's met all four criteria (50+ events, 5+ min span, 3+ distinct paths, no valid human marker). Escalates 1h → 6h → 24h across separate offenses; permanent only ever happens manually.

🔌 Redis outage history

PoL verification itself no longer depends on Redis (signed, stateless), so an outage here no longer means rejected visitors - this is visibility into the arm/disarm engine's health, not an active-incident panel to panic over.

🕘 Audit log

Every mode/config change to a rule, with who made it, when, and the exact old/new values - written in the same transaction as the change itself, so this can never drift from what actually happened.

No Website Selected

Select a website to manage its configurations, trigger publishing tasks, inspect feeds, and review diagnostics.