Skip to main content

Integration with ChatMCP

How to set up ChatMCP for neural network access via BotHub​

Note. The Kaas program has limited features. If you need more flexibility, it's better to choose one of the following solutions:

• Msty, Cherry Studio (for Windows, Linux, macOS);

• Open WebUI (server + browser access).

ChatMCP is a program for accessing neural networks for Windows, Linux, and macOS.

🛠️ How to install​

Download installers for Windows, Linux, macOS from the GitHub releases, also an .apk file for Android.

For iOS, installation is done via TestFlight.

⚙️ How to connect and what to configure​

  1. Go to Settings → Providers.

  2. Click Add provider, enter “BotHub”.

  3. Set parameters:

  1. Click Fetch to load the list of models.

  2. In the Gen title section, select a model that will generate dialog titles. It’s better to choose a lightweight one, such as qwq-32b or gpt‑oss-20b.

  1. Finally, click Save settings.

Tip: In General → System prompt, shorten the system prompt, leaving only the line "Use Mermaid to generate diagrams." All other phrases usually do not affect the model's behavior.

🧙 How to use ChatMCP​

ChatMCP looks like a standard LLM client: at the top, you select a model; below, there’s an input field and a send button.

File support in ChatMCP is very limited: you can only attach images (one at a time) and source code files.

The new chat button is located at the bottom of the dialog list and to the left of the minimize button.

The current model parameters — t°, top‑P, max tokens, frequency penalty, presence penalty — are in the ⋮ menu → Model config.

More detailed article: https://habr.com/ru/companies/bothub/articles/953986