EDAi Settings Dialogue Form
Overview
The ProPilot Configuration dialogue is where you set up access to your AI model, describe the project you are working on, configure external chat client access, and set the appearance of the chat panel. Open it with the cog button at the top right of the chat panel.
The dialogue has four tabs:
- Project Settings, where you tell the assistant about the current project and, if you are an educator, manage the lesson plan
- LLM Provider, where you choose a model provider, enter your API key and set the model options
- MCP Server, where you enable access from an external chat client
- Font, where you set the chat panel font
Changes are applied when you click OK and discarded if you click Cancel.
Project Settings Tab
This tab holds everything that belongs to the current project rather than to your installation. It opens first, because it is the tab you are most likely to want.
Project Specific Instructions
Anything you want the assistant to understand about the current project. This is often unnecessary, but a top-level description gives the model useful background and helps it produce answers that fit the design. For example:
"This is a GSM-enabled IoT monitoring and control device that uses an STM32F401CE microcontroller with an IMU sensor to monitor conditions and communicate wirelessly via a SIM800C GSM module. The device can remotely control external equipment through a relay output while monitoring 12V power systems and providing USB connectivity for programming and data transfer."
When Tutor Mode is enabled, this field is relabelled Lesson Plan and takes on a second role, described below.
Tutor Mode
Switches the project from ProPilot to ProTutor, so that the assistant teaches rather than answers. When enabled, the assistant uses the lesson plan and the user instructions configured on this tab. Ticking the box also reveals the lesson plan controls.
Unticking Tutor Mode discards the lesson plan held in the project. There is no confirmation prompt and the plan is removed when you click OK. If you want to stop editing a lesson but keep it, untick Authoring instead and leave Tutor Mode ticked.
Authoring
Makes the lesson plan editable so that you can build or revise it in conversation with the assistant. Authoring implies Tutor Mode, so ticking this box ticks Tutor Mode too. The box becomes available once the project contains a lesson plan.
With Authoring off, the plan is sealed for running and the assistant delivers it as a tutor. With Authoring on, the assistant works with you to design and rehearse it instead.
Lesson Plan
With Tutor Mode enabled, this field shows a summary of the lesson plan currently held in the project: the title, description, author, duration, target audience, whether auto-marking is enabled, and the list of steps with their types. The summary is read-only, since the plan is edited through the assistant rather than by hand.
For a simple fault-finding exercise that does not need structured validation, you can instead type or paste plain text instructions into this field. For example:
"There is a deliberately created fault in this circuit. U1 is short circuit at +IP and -IP input pins. The student's task is to identify which component is faulty. Your role is to help them by suggesting measurements to take. The student can measure voltages and currents by placing probes and then simulating the circuit."
User Instructions
The text shown to the student at the top of the chat window when they open ProTutor. Use it to set the scene for the exercise. For example:
"The circuit is supposed to be a non-inverting amplifier but there is no signal at the output. One of the components is faulty. You need to work out which one. ProTutor will help you find the fault."
This field appears for free-text exercises. A structured lesson plan carries its own opening instructions, so the field is not shown when a plan is loaded.
New Lesson Plan...
Creates an empty structured lesson plan. Proteus asks for the lesson title, author, description, target audience and expected duration, and whether results should be shown to the student at the end of the session. Creating a plan also turns on Authoring, ready for you to design the lesson.
Import Lesson Plan...
Opens a file browser to select a lesson plan JSON file. The file is validated and its contents stored inside the Proteus project file. A file that is not a lesson plan is rejected with an explanation.
Export Lesson Plan...
Saves the current lesson plan to a JSON file, suggesting a filename based on the lesson title. Useful for sharing with colleagues, keeping a version-controlled copy outside the project file, or recovering a plan later.
Edit...
Reopens the lesson details supplied when the plan was created, so that you can revise the title, description, audience, duration or auto-marking setting.
Lock and Unlock
Locks the lesson plan with a password so that students cannot view, edit or export it. While the project is locked, the plan is hidden in this dialogue and the lesson plan buttons are withdrawn. Click Unlock and supply the password to work on the plan again.
Locking is tamper evidence rather than encryption: the password fingerprint covers the plan text, so a plan altered outside Proteus will no longer unlock. The plan text itself remains within the project file.
See Also:
LLM Provider Tab
This tab configures access to your chosen large language model provider. It is typically a one-time setup, revisited when you change provider or move to a newer model.
Provider
Choose from the supported providers: Anthropic (Claude), Open AI (ChatGPT), DeepSeek, Google (Gemini), Moonshot (Kimi), Ollama (local) and Custom. Selecting a provider fills in the other fields with the appropriate defaults, and settings you have entered for a provider are remembered if you switch away and back.
API Key
The API key issued by your provider. This is required for authentication and usage tracking. The field is masked, and the key is stored locally on your machine. The label changes to name the provider whose key is wanted, so you can see at a glance which account the field belongs to. Ollama needs no key.
Base URL
The endpoint Proteus sends requests to. This is filled in automatically and cannot be changed for the mainstream cloud providers, since there is only one correct value. It is editable for Ollama, for Custom, and for Moonshot (Kimi), where users in China need to point Proteus at the corresponding regional endpoint.
Model
The model to use. Click Refresh to fetch the list of models available on your account from the provider. The list is not fetched for Custom endpoints, so type the model identifier in yourself.
Refresh
Loads the latest model list from the provider using the API key you have entered. Run it periodically, since new models appear in the list only when you refresh it.
Temperature
Controls how much variation there is in the model's responses. Lower values, for example 0.2, make answers more deterministic; higher values encourage more variation. A typical range is 0.0 to 1.0. For engineering work the lower half of the range is usually the better choice.
Note that some models ignore this setting when Extra Thinking is enabled, and that a few models reason deeply by default and do not accept it at all.
Max Completion Tokens
The maximum number of tokens the model may use for a single response. This limits the length of the reply only; it does not limit the total context, which also includes the input.
Test Connection
Sends a short test request to verify that your API key, base URL and model selection work together. Success is confirmed with a message; failure reports what went wrong, so that you can tell an invalid key from a network problem. If no key has been entered, Proteus says so rather than attempting the request.
Enable Web Search
Allows the model to search the live internet rather than relying only on its trained knowledge. This is supported on Anthropic and Open AI models; the checkbox is greyed out on providers that do not support it. Even where it is supported, models tend to answer from trained knowledge by default, so you will often need to ask explicitly for a live search.
Enable Extra Thinking
Uses the model's deeper reasoning mode where available; the exact behaviour varies by model. It can improve results on reasoning tasks such as multi-step calculations and design analysis, and it consumes more tokens. Supported on Anthropic and Open AI models; the checkbox is greyed out on providers that do not support it.
Enable Streaming Mode
Shows the response as it is produced, along with progress feedback at the foot of the chat window, rather than waiting for the complete reply. Supported on Anthropic and Open AI models; the checkbox is greyed out on providers that do not support it.
See Also:
MCP Server Tab
This tab lets an external desktop chat application drive Proteus. All of it is off by default.
MCP Server Port
The port the server listens on. The default is 8001, and the same port must be configured in your chat client.
Enable MCP Server
Starts the server. Clear the checkbox to stop it again.
Allow file operations to discard unsaved changes
Permits an external assistant to open a project or close Proteus while the current design has unsaved changes. With the option off, those operations are refused and the situation is reported back to the assistant instead, so it cannot discard work you have not saved.
Enable schematic drawing tools (beta)
Exposes the experimental schematic drawing tools, which allow an assistant to place devices and terminals, route nets and edit the sheet. These are under active development and are not yet fully functional. The rest of the schematic toolkit is unaffected by this setting.
Single element results
A compatibility mode for chat clients that read only the first element of a tool result. It collapses each result into a single element so that nothing is silently dropped. Leave it off unless your client needs it.
See Also:
Font Tab
This tab sets the font used in the chat panel: family, style and size. It exists for readability. Chat text is dense, and a larger or clearer face makes a considerable difference over a long session, particularly on a high-resolution display. A sample of the chosen font is shown as you make the selection, and the setting is remembered between sessions.
The font applies to the chat panel only and has no effect on the schematic, the code editor, or anything else in Proteus. For quick adjustments while you are reading, you can also hold CTRL and turn the mouse wheel over the chat, then use the zoom button to return to 100%.
See Also: