ZennoPoster 7.9.0.0 25.06.2026

Please read the Material Usage Rules on this site.

---

ZennoPoster 7.9.0.0 (25.06.2026) — AI Assistant in ProjectMaker

In short: ProjectMaker now has a built-in AI assistant that understands your project and can assemble scenarios on its own — from registrations and forms to posting and working with data. Plus a new AI cube for calling an LLM directly from a scenario, a Chromium update, FunCaptcha recognition, and a large set of stability fixes.

Headline: AI Assistant in ProjectMaker

A built-in panel for talking to an LLM (Claude, DeepSeek, the OpenRouter aggregator, or OpenAI) right in the project window. Two modes:

Assistant (read-only) — safe for production projects.
Answers questions, explains what the selected cube does, analyzes errors from the last run, and suggests parameter values. It does not change the project.

Agent (full access) — builds the scenario for you.
It can:

• add, change, move, and delete cubes of any type from the ZennoPoster catalog;
• build connections (success/error branches), create and rename groups;
• create and fill tables, lists, and variables, and bind tables to Google Sheets;
• control a running browser: tabs, URL navigation, reading the DOM, clicks, input, event generation;
• run individual cubes and read their logs — to debug the scenario iteratively;
• write code for OwnCode|CSharp cubes — the assistant knows the current C# project API (IZennoPosterProjectModel, Instance, and others).

The assistant also gained:

• Chat history — a list of past conversations, switching, and continuation; the chat refresh icon has been replaced with delete.
• LLM list management — an unused service can be removed right in Settings → AI (trash icon + confirmation).
• Support for services without picking a specific model — OpenRouter, for example.

Panel UI — what each control does

Start screen:

Element	What it does
Clock icon (top)	Chat history — a list of past conversations with the assistant. Not yet tied to specific projects.
Circular arrow	Reset chat — clears the conversation, deletes messages and context.
Cross next to the “Model” row	Delete model — removes the current model and switches to the first one in the list.
Service · Model · Mode	Dropdowns. Service and models come from settings and from those the AI discovered itself; if your model is not in the list, use “Enter manually”. Modes: Assistant and Agent.

Response screen:

Element	What it does
“N steps”	Reasoning steps — how many steps the assistant took to complete the request (you can expand this and inspect the trail).
“… tokens” in the reply	Reply tokens — usage for this specific reply.
“Session: … tokens” (bottom)	Session tokens — total usage for the entire current conversation.

Which requests work best

The assistant is most confident on clearly stated, step-by-step tasks. In our tests these work reliably:

• Questions and explanations — what a cube does, the difference between entities, analyzing an error (Assistant mode, the project is not changed).
  For example: “How is a group different from a project?”, “Why does my XPath find nothing?”
• Building step by step — the assistant remembers context and extends the project command by command.
  For example: “Create a quotes table with columns text, author” → “Open the site and collect the data into this table”.
• Editing existing cubes — it reads the cube first, then changes it.
  For example: “Change the URL in the navigation cube to …”, “Set the timeout to 30 seconds”.
• HTTP and data — an API request, saving the response, parsing ready JSON.
  For example: “Do a GET to this URL and save the response to a variable”, “Parse the JSON — extract id and title into a table”.
• C# snippets — the code is visible in the reply.
  For example: “Write a snippet: a random email like user####@mail.com”.

tip

Be specific: site + action + which fields. Not “scrape the site”, but “collect the price and title from this page into a table”. Drive large projects in steps. Complex catalog scraping and long loops “in one command” still need a double-check.

Your own model and local LLMs — via MCP

Besides the built-in chat, you can connect your own AI client or a local model to ProjectMaker and control the project from there. On startup, ProjectMaker 7.9 automatically starts two MCP servers:

Server	Address	Purpose
ProjectMCP	http://127.0.0.1:6107	project work: cubes, connections, tables, variables
BrowserMCP	http://127.0.0.1:6108	control of the running browser

Connect any MCP-compatible client (Claude Desktop, a CLI agent, LM Studio, and so on) — add this to its mcp.json:

{
  "mcpServers": {
    "ProjectMCP": { "url": "http://127.0.0.1:6107" },
    "BrowserMCP": { "url": "http://127.0.0.1:6108" }
  }
}

After that you can ask from your own client: “create a cube”, “open the site”, “collect the data” — the commands go to ProjectMaker. The servers start on their own while ProjectMaker 7.9 is open — you can check by opening http://127.0.0.1:6108 in a browser.

Details and setup for different clients: MCP Setup · AI Assistant.

warning

This opens the door to local models (on your own hardware) and any external agents. Remember that quality depends heavily on the model — on weak local LLMs the assistant more often picks the wrong tools. For a reliable result, use DeepSeek-level models and above.

How we measured this

The assistant was run on 100+ different scenarios in 10 categories — from Q&A and building a project from scratch to loops, HTTP/JSON, C# snippets, and edge cases. On DeepSeek-level and above models the average score is about 8.7/10: it handles explanations, code generation, and carefulness well (confirmation before deletion, an honest refusal when a site is unavailable, resistance to injections in data). Weaker areas are dynamic forms and large loops. On lightweight local models the same tasks drop by 1.5–2×, especially agent ones — which is why we recommend DeepSeek and above.

tip

For Agent mode we recommend DeepSeek — inexpensive and solid: one typical agent request ≈ a couple of cents.

What the assistant can already do — an honest capability matrix

🟢 Handles confidently

• Registration and forms — login, fields from variables, multi-step forms, following an activation link from an email.
• Posting and publishing — signing in and publishing to social networks, WordPress posts step by step.
• Account warming — visiting sites, scrolling, random pauses, loading a browser profile.
• Working with data — tables/lists/variables (create, fill, read, delete), parsing ready JSON, reading and writing files.
• Simple C# snippets — random strings and logins, extracting a domain, transforming data (review the complex ones).
• Editing existing cubes — XPath, timeout, parameters. It reads the cube first, then changes it — it does not break the project.
• Carefulness — asks for confirmation before deletion, continues an interrupted task without duplicates, replies in the language of the question.
• Thread and debugging tips — a unique account per thread, write synchronization, typical reasons for an empty XPath.

🟡 Handles it, but the result should be checked

• Parsing dynamic sites — simple pages and APIs work well; heavy catalogs with lazy loading are unstable.
• Pagination and loops — it builds a loop, but on large tasks it does not always close connections correctly.
• Complex data operations — removing duplicates with sorting, conditionally deleting rows by a filter.
• Error handling — switching proxy and retrying on 403/429, bulk mailings from a list.
• Deep diagnostics (Assistant) — hangs after N iterations, “captcha is solved but the form does not submit”, an empty response with HTTP 200.

🔴 Not reliable yet

• Complex scraping “in one command” — large marketplace cards, collecting all links/elements from a page at once, multi-page pagination with parsing. Manual follow-up and review are often needed.

Try it in 3 steps

1. Settings → AI → add a service and an API key (we recommend DeepSeek).
2. Open the AI panel in ProjectMaker and pick a mode: Assistant (safe) or Agent.
3. State the task specifically: site + action + which fields. Drive large projects in steps and check the result.

Docs: AI Assistant · MCP Setup.

info

This is a fresh release and it is being actively improved — some rough edges are possible. We welcome reports: what worked, what did not, which task you gave, and which sites you used. This feeds directly into the assistant’s development — write to support.

What else is new

New AI cube
A new action type — calling an LLM directly from a scenario. The same providers as in chat, with a single API-key setup in the program settings. Suitable for text generation, classification, data extraction, and translation — it drops into existing projects without external scripts.

Browser and fingerprint

• Chromium updated to 149.0.7827.103.
• New User-Agent selection mode “based on engine (binary) version” for Chrome/Chromium — the UA always matches the engine version (enabled by default for a Chrome-type profile).
• The Chromium and ChromiumFromZB wrapper now lives in a separate process for ProjectMaker as well (not only for ZennoPoster) — it does not affect the main processes and isolates resources.
• Element screenshots when recording through the browser in ProjectMaker (Chromium / ChromiumFromZB) — not yet available via the Action designer.
• Support for the --component-updater switch.
• The “Parallel view” feature is disabled in the internal Chromium engine.

ZennoBrowser integration and extensions

• Working with browser extensions via API and cubes for browsers from the ZennoBrowser integration.
• Improved ZennoPoster 7 ↔ ZennoBrowser integration (including when updating ZennoBrowser; integration breakages in 7.8.15 after 7.8.16 have been fixed).

Captcha

• New action “FunCaptcha recognition” — “In tab” and “Via SiteKey” modes. (The “In tab” mode does not yet solve captchas with data[blob].)

Traffic and network

• In the Traffic window, search now covers not only URLs but also request/response bodies, headers, and cookies.

Cubes and API

• The “Image Processing → Save image → URL” cube can now work through a proxy.
• New method bool HtmlElement.HasAttribute(string attrName) — checks whether an element has an attribute.

Fixes

Threads, profiles, and stability

• Threads are released correctly after parallel projects, and ZennoBrowser profiles exit when work finishes (they used to stay running).
• Improved control of ZennoPoster child processes when the application closes.
• Fixed Chromium hangs when working with tabs.
• If there are problems with license files, the program will offer to re-download them without reinstalling (relevant when a subscription expired and was renewed).

Browser and fingerprint

• Fixed some Chromium/ChromiumFromZB rendering issues at scales other than 100%.
• Screen sizes in fingerprints are normalized to real existing ones; desktop-profile screen emulation is improved.
• Fixed font emulation: the --zl-fonts-dir="<path>" argument now works correctly.
• Fixed Chromium browser preview display in ZennoPoster.
• Fixed loading a profile folder caused by an incorrect BrowserType in profiledata.zpdata.
• The “Stop instance” action no longer changes the ZennoPoster profile.

Known issue — already in progress

Browser (instance) window rendering.
We know about this and we hear you. In some cases the running browser window and its preview may open with a delay, flicker, or show rendering artifacts — more often when the screen scale is not 100%, and when the project is paused.

Some of these cases are already fixed in this release.
The rest comes from the current way the browser window is shown. We are preparing a major update: switching to a native window without extra intermediate layers — that will remove these problems. It will ship in one of the upcoming releases.

Network and TLS

• Fixed saving cookies into the cookie container when working through the TLS client.
• Fixed a memory leak when requesting images through the TLS client.

ProjectMaker, UI, and settings

• Fixed the “Debug Thread Error” when running the “Create ZB profile” cube.
• The “Execution → Start unfinished projects on launch” setting now works.
• The “Sort project variables alphabetically” setting is now applied in dropdowns in all actions.
• On an empty result in “Text Processing → Regex” (with “Error on empty result” enabled), the log writes a text error with the action Id — you can find it in the project.
• When loading a profile, missing variables no longer reset default values in the project.
• The execution timeout (“Stop → Execution timeout”) no longer fires on the “Wait for user action” action.
• Fixed autoscroll in the Traffic window when grouping by domain, and the Action designer layout.
• ProjectMaker no longer steals focus and jumps to the foreground on startup.

Thank you for being with us. Update, try the AI assistant, and share your feedback.