Are MCP and AI preferences completely broken?
For the past six months i’ve been using BoltAI, MCP and AI preferences have not worked for me. Yes, I have read all the posts and the web.
New project chats do not inherit MCP server or default AI settings.
New Projects do do not inherit MCP server or default AI settings.
Preferences seem useless.
I have to set all MCP server settings and the AI with every chat spawned to a project row. I have to set set all MCP server settings and the AI with every new Project.
Also, there are three separate places to enter AI settings. Settings are a hot mess.
Log in to comment and vote
Comments6
Nigel Protter
Sep 5
Mac OS 26.6.2, BoltAI v2.15.0 B84
Daniel Nguyen
Sep 5
Hey Nigel.
Sorry this has been frustrating, especially when you’ve already spent time reading the documentation. I agree the settings UI could make it much clearer which configuration applies.
The multiple settings screens serve different scopes. A chat uses its selected profile:
Customized: the chat’s own settings.
Project: the project’s settings.
Agent: the selected agent’s settings.
Default / Base Agent: your default configuration.
These aren’t all applied together. For example, a chat using a customized or agent profile won’t use the project’s configuration.
That may explain some of what you’re seeing, but I don’t want to assume it explains everything. Could you share your BoltAI version and one specific example:
Where did you set the AI model and MCP selections?
How did you create the new project or chat?
Which profile was selected in that chat?
What settings did you expect, and what appeared instead?
A short screen recording would also help. With those details, I can follow the same steps and determine whether this is a bug or something the UI needs to explain better.
Thank you
Nigel Protter
Sep 5
Daniel, What I think a typical user would expect is that a chat spawned from a project inherits that project settings. Why wouldn’t you want that as a default, rather than no settings?
Similarly, a new project should inherit the settings in the preferences as a default. Again, why wouldn’t anyone expect that?
Third, the settings in preferences are confusing to begin with, for example “Base Agent” is always visually at the top of the list, even if it isn’t the user’s preferred default. Thats is a visual usability contradiction. Just get rid of base agenht completely and move the preferred agent to the top of the list.
Then there are agent settings in three places, some link to each other, I understand that, but AI settings lets you adjust more agent settings than project AI settings, for the exact same agent, why? Chat context adds a third wrinkle that confuses, I realize its linked, but what is it adding? why have it at all?
Settings should cascade down, never sideways or up, then let users customize at the project or chat level as desired.
If there is a logic to settings to do what I want that I am missing, (and I assume there is!), then I feel it’s too complicated, not intuitive.
Daniel Nguyen
Sep 5
Nigel, to address the main point directly: a new chat inside a project does start with that project’s settings. Your recording shows the project’s model carrying over to the new chat.
The confusing part is that the selected tools carry over separately from the master switch that enables tool use. That switch is off in the new chat you show. The agent picker also doesn’t clearly indicate that the chat is using project settings. I understand why this looks like the settings haven’t carried over, and I’ll improve the wording and help text.
From your recording and explanation, are you expecting your selected default agent to supply the settings for new chats inside projects too?
If so, that’s the distinction: the default agent applies to new chats outside projects. Chats created inside a project start with the project’s configuration. Otherwise, choosing a default agent would override the settings you deliberately configured for that project.
The project settings screen therefore isn’t another editor for the same agent, even if both use the same model. They are separate configurations.
This is a tricky UX balance: some people want one default everywhere, while others configure different projects for different work. BoltAI gives project settings priority within that project. That behavior is intentional, but the interface needs to make it much clearer, which settings apply, where they come from, and whether tools are enabled.
For your setup, open Settings → Chat Context, temporarily select Base Agent, then open Manage Plugins and turn on Enable tools by default. You can then restore your preferred default agent. After that, configure the model and tool selections in the project’s AI Model Settings and Manage Plugins; you don’t need to change the default agent’s configuration for project chats. New chats in the project will use those project settings with tools enabled. For the existing chat in your recording, turn on the master switch in its tools menu.
Thanks
Nigel Protter
Sep 6
Will the system instructions pasted into the project will have an effect in chats spawned from a project row? ? Or do i have to paste them into “default Chat Context” system prompt?
Daniel Nguyen
Sep 6
Yes, the system instructions you enter in the project apply to new chats created from that project’s row. Keep the chat on its Project Profile to use those instructions.
You don’t need to copy them into “Default Chat Context.” That system prompt is for chats using the Base Agent outside projects.