A small, nerdy premise.
If you notice a few nods to Ghostbusters (1984) along the way, that’s no coincidence.
WebWall v1.2 introduces three distinct roles that give structure and logic to its New Architecture:

– The Gatekeeper – stands at the door.
– Gozer the Destroyer – decides whether the structure holds.
– The Keymaster – maintains order and balance.

The naming is playful, yes—but the roles are very real, and each exists for a reason…

Originally, WebWall was a simple, efficient bouncer. It didn’t analyze or philosophize, it just worked:

“You: yes.”
“You: no.”

For a time, that simplicity was enough. Requests arrived, rules were checked, and anything suspicious stayed out. Fast. Clean. Predictable. But the web changes and today, abuse often isn’t a single bad request. It’s pressure over time. This is why, an important upgrade on WebWall Core was needed.

Blocking a request in 2026 is easy. Knowing when to stop forgiving is not.
That’s why v1.2 changes everything. WebWall now looks beyond the present: it observes patterns, behavior, and time. It remembers, but only just enough.

The Gatekeeper

The Gatekeeper lives in the request lifecycle. It sees everything but decides very little.

It does:

– Track visitors and update hit counters
– Apply lightweight scoring
– Enforce immediate blocks when thresholds are exceeded

It does not:

– Decide permanent bans
– Perform heavy logic
– Talk to external services

Think of it this way: the Gatekeeper keeps the door open, ensuring access control without unnecessary judgment. Most visitors will meet only him, and that’s perfectly fine.

Introducing Gozer

Gozer doesn’t run on every request. It doesn’t care about single hits: it cares about behavior.
Running asynchronously, Gozer looks at the bigger picture and “chooses the form of the Destroyer“:

– Accumulated scores
– Repeated abuse
– Pressure over time
– Decay of inactivity

It answers one question:

“Has this visitor earned consequences?”

There’s no machine learning, AI, or black-box guessing. No fingerprinting. No magic.
Gozer relies on transparent logic:

– Scores rise with bad behavior
– Decay rewards inactivity
– Escalation happens only if pressure persists

If a visitor stops, their score fades. If they insist, Gozer remembers. Persistence is expensive.
From the admin’s perspective, this process translates to:

– Pulse: active tracking
– Temporary blocks: suspicious behavior
– Blacklist (Burned): permanent bans
– Honeypot IPs: known hostiles

Behind the scenes? For esample:

– Score > 100+ = permanent local ban
– Continued activity = escalation
– Extreme abuse = Cloudflare-level block

At that point, the visitor never reaches the server again and the Gatekeeper doesn’t even see them.

The Keymaster

The Keymaster operates in the shadows, within the backend.
He maintains communication, synchronizes data, and keeps systems aligned.

His duties:

– Update the blacklist with bad IPs
– Maintain whitelists for trusted engines
– Check blacklists via lightning DNS
– Push escalation rules to Cloudflare instantly

Think of him as Gozer’s deputy, holding the keys that preserve balance and stability across layers.

A day in the life

When a domain goes live with WebWall v1.2, here’s what a typical 24-hour cycle might look like:

Hours 0–2:
Crawlers and scanners arrive. Scores stay low. The system observes silently.

Hours 3–6:
Aggressive probing begins. Temporary blocks appear.

Hours 6–12:
One IP keeps pushing. It crosses the line. Gozer burns it.

Hours 12–24:
Persistent hits trigger escalation. Cloudflare blocks them at the edge.
Within seconds……….silence: no manual action required.

No alerts. No babysitting. Just logic doing its job.
The separation of Gatekeeper, Gozer, and Keymaster isn’t cosmetic: it’s architectural.

• The fast path remains fast
• Heavy logic runs asynchronously
• Decisions grow smarter over time
• Updates adapt to new threats seamlessly
• The Gatekeeper says: “You may enter…”
• Gozer eventually says: “You shouldn’t have…”
• The Keymaster whispers: “I know you all…”

So, zero dependencies, zero drama
WebWall v1.2 runs wherever you drop it.

• No frameworks
• No external services
• SQLite for live state, MySQL for history
• Optional Cloudflare escalation

Setup is intentionally… boring, and that’s a compliment.
Final note? Yes, Gozer has a link. And yes, it points here:
https://ghostbusters.fandom.com/wiki/Gozer

Because being a little nerdy never hurt anyone, especially when the system just works-and this one does.
PS: There’s one more character worth mentioning, the funniest one, “Slimer“…and indeed, there is. He only does one thing besides making people smile: he tells Gozer who’s bothering him. And if he does…well, we already know what happens.

Discover WebWall 1.2
Try the Live Demo

 

Imagine you’re on a Forum you’ve never seen before.
Threads on Cooking, Tech, Books, side projects… people chatting, joking, helping each other.
Among them—hidden in plain sight—is a User called “Dokky“.
Profile picture, Bio, Forum Badges. Nothing screams “Bot“. You scroll, read a bit, until you notice something strange: Dokky replies to everyone precisely, kindly, often in their own language. And when someone writes, “Hey @Dokky, what do you think?“, he arrives, right on time, inside the thread. Not in a popup. Not in a separate chat. Right there, where the conversation lives.

Welcome to the Dokky Suite 2.0.4: a Document Publishing and Community Platform that has decided to integrate AI not as a chatbot… but as a real forum member.

Most AI integrations on websites today look the same: a bubble in the corner, a chat window, a “How can I help you?” prompt. Useful? Sure. But isolated. The conversation is usually 1‑to‑1, detached from what the rest of the community is doing. With Dokky 2.0.4, the Idea is the opposite: let the AI step into the shared space, into the threads, and become part of the Social fabric.

There is no “bot area” here. There is a user.
The Community calls them by name – “@Dokky” – and the AI answers in the same place where everyone else is talking. It reads the original Post, scrolls through previous replies, senses the mood and the Topic, and then joins the discussion as one more voice. Not the main voice. Not the official voice. Just another presence… with a very fast brain.

For example, in a cooking Forum, this presence can feel like the wise aunt who never runs out of tricks.
Someone posts a recipe for lasagna; a few replies later, a user tags @Dokky:

“Any idea how to save this if I overcooked the sauce?”

Dokky appears with a suggestion: how to adjust seasoning, how to rebalance textures, maybe even a little cultural note about how this dish is done in another region or Country. No long lecture, no condescending tone. Just a practical, friendly tip, perfectly in context. A few days later, the same user asks about gluten‑free alternatives; Dokky is there again, acting like an old friend, a patient companion who remembers the conversation and continues it.

In a Tech Community, the same “entity” can wear a completely different mask. The admin might configure Dokky as a witty but precise “resident nerd”: short answers, clear explanations, examples when needed. A newcomer struggles with a concept in APIs or caching? Instead of digging through documentation, they tag @Dokky in a thread where other members are already debating. Dokky joins in, not as a support agent, but as someone who “hangs out there” and knows their stuff.

Or again, over in a comic‑book or Bookstore Forum, Dokky can become the Librarian you wish you had when you were a kid. Someone posts about a Graphic Novel they loved; they ask “What else should I read?” and tag @Dokky. The AI replies with three or four well‑chosen Titles, one slightly mainstream, one niche, one classic. Maybe it adds a short note: “If you liked the way this story plays with timelines, you might enjoy…” It doesn’t feel like a recommendation engine. It feels like a person who knows the shelves by heart and actually listens.

What makes this possible isn’t just “using an AI model” – that part is easy today. The real trick is giving Admins a way to shape the personality and role of this AI user without writing a line of code.

Inside Dokky 2.0.4, there’s a dedicated area where the administrator can:

– Pick a Model from a curated list (10 Different AI Models, some free, some paid), and switch later if the community grows or needs change;
– Choose a base “preset” for the AI’s persona – more playful, more technical, more culinary, more bookish – and then refine it in natural language, like writing a character brief;
– Decide how often the AI is allowed to speak, how much of the previous conversation it can see, how creative it should be, and how long its replies should roughly be.

It’s a bit like casting a new Member into your Community and then giving them some guidelines: “You’re friendly but not cheesy”, “You’re technical but patient”, “You’re here to help, not to dominate”. Over time, the Admin can tweak these traits: less verbose, more direct, more empathetic, more fun. Every Dokky installation can grow its own “Local Legend” – a unique AI user that matches the culture of that specific Forum.

There’s another subtle, important dimension: rhythm.
Whenever you introduce Automation into a Social Space, there’s a risk that it becomes too loud, too present, too fast. Dokky avoids this by playing by the same rules as everyone else: it only speaks when mentioned, it respects limits set by the Admin, and under the hood there’s a gentle throttling mechanism that prevents it from answering too many calls in a short time. The feeling, for Human Users, is not “the bot is everywhere”, but “Dokky shows up when we actually call for it”.

This balance is what turns the idea from a gimmick into something sustainable. In a long thread, where no one has the energy to recap everything, Dokky can be the one who remembers the whole story. In a multilingual Community, it can smoothly shift language to include someone who otherwise might feel left out. In a quiet Forum, it can throw a question back to the Group – “What do you think about this new feature?” – and help kickstart more human replies.

All of this lives inside a larger evolution. Dokky started life as a Powerful Script for Managing and Publishing Documents: Uploads, Protection, Licensing, Monetization, Forums, Private Groups, all the heavy engineering you’d expect from a modern PHP suite. Over time, releases have focused on performance, caching, GDPR, analytics, OCR, IPFS, ads, and compatibility with the latest PHP versions. The machine became solid first.

With version 2.0.4, the focus shifts from “what the Platform can do” to “how the Community feels while using it”. An AI user is not a feature checkbox; it’s a statement of intent: this is a place where automation and humans share the same space, and where the goal is richer, more inclusive, more alive conversations.

So if you’re looking for “AI + community” and you’re tired of yet another chat bubble that nobody clicks on, Dokky’s approach is different: give your Forum a new voice – one you can design, refine, and grow – and let it sit at the same table as everyone else. Not above them. Not outside. Right there, in the thread, ready to answer when someone simply says:

“Hey @Dokky, what do you think?”

Live Demo
Get Dokky Suite 2.0.4

 

If there’s one software that never stops evolving, it’s Dokky. From version 2.0.0 to 2.0.3, the leap is remarkable: many concrete improvements, no frills, and a clear philosophy: to make Document Management truly independent and user-friendly.

Among the features that truly change users’ lives is the new “dynamic chunk” Upload system. Essentially, Dokky breaks files into parts and adjusts their size based on their weight: this way, the platform quickly uploads small attachments and doesn’t crash when faced with giant files. This truly optimizes resources; less risk of errors, less stress for servers and users. The upload UI has also been redesigned, with a progress bar and visual support for synchronization on the IPFS Network, for a modern and dynamic experience.
Another new feature is the creation of short URLs for Documentation and Posts, or native QR codes without plugins or external tools.
An OCR service has been integrated for Documents. This powerful feature allows you to semantically extract part of the Document’s content and use it to automatically and precisely populate Titles, Descriptions, and meta tags.

Another significant new feature is the refactoring of the ADS system. Dokky now has five standard “format” positions for displaying ads. It’s a balanced choice: monetise, but don’t be intrusive. Ad management now supports multiple formats, randomization, and compatibility with Google Adsense. Even the new API admin section is designed for clarity: everything under control, no time wasted on scattered pages and panels. But there are even more interesting new features. User privacy is a critical and central issue today: the new GDPR cookie consent system makes the difference. It’s not just about “showing the banner” for the sake of form. Dokky truly blocks cookies until consent is given, and tracks them anonymously (with an anonymous log and the ability to download the CSV of accepted/denied cookies). This allows admins and users to work with peace of mind: transparency and legality are just a click away.

Small details, great usability! Behind every improvement lies attention to detail. From the new Google Analytics compatibility to the OCR integration for meta-documentation management, to the forum badges on public profiles, every feature stems from real requests and provides concrete value to those who use Dokky every day. Whether you’re an experienced developer or a novice user, the goal is always the same: to work better, with less stress and more autonomy. Dokky continues to grow “with you”, listening to the Community and focusing on what really matters.

Want to discover all the new features and try them on your Dokky instance? Just one click: the Documentation is updated and our Team is ready to answer any questions! Do you have other ideas or needs to share? Contact us and help us build the Document Suite of the future!