User Guide#

Use this guide as the main entry point for working with Simplicity AI SDK after installation. It connects the setup, authentication, project, and responsible-use topics into a typical development flow.

VS Code Extension#

Install the Silicon Labs Support Extension as part of your platform setup. See Installation Overview and the OS-specific guides for Windows, Mac, and Linux.

Prerequisites#

Before you use Simplicity AI SDK workflows, confirm the following:

SimplicityInstaller completed with MCP packages installed
Simplicity SDK at the version required by your release
Visual Studio Code with an AI agent enabled and a capable model selected
A supported host OS (Windows, macOS, or Linux)
For hardware workflows: a supported Silicon Labs board connected through a debug adapter
For documentation and Silicon Labs knowledge base access: a Silicon Labs Community account and completed OAuth Authentication Process

Complete Installation Overview before you run build, flash, or debug prompts.

Supported Workflows#

Create new project - Generate an example from a prompt, then modify, build, flash, and validate it. See AI-Assisted Project.
Import and modify existing project - Open a Simplicity Studio-generated project in your IDE and extend it with the agent. See Work with Existing Projects.
Debug existing project - Investigate runtime faults, incorrect behavior, or assert failures on a connected board with agent-assisted logging and debug sessions. See AI-Assisted Project.

Recommended Development Workflow#

Discover -> Use silabs-knowledge-base and slab-example-find to locate example projects, documentation, compatibility guidance, and implementation patterns.
Connect hardware -> Attach your target board and confirm the agent can detect adapters.
Create new project or open existing project -> Open an existing project or use slab-example-create via silabs-tool-locator to create a project from an SDK example.
Generate -> Use slab-project-generate via silabs-tool-locator to generate the project.
Configure -> Use slab-project-modify and slab-cmsis-config-edit to add components and set project configuration, then regenerate with slab-project-regenerate.
Modify -> Describe application changes in natural language. Use silabs-knowledge-base when you need Silicon Labs APIs, components, or examples. Implement changes incrementally.
Build -> Compile with silabs-build-tool or silabs-vscode-extension.
Flash -> Deploy firmware to the target with slab-sdm-hardware.
Debug -> Use silabs-gdb-mcp or silabs-vscode-extension for interactive debug, and slab-sdm-hardware for serial output.
Test -> Use slab-bgapi-cli for NCP workflows, slab-sdm-hardware for capture from the adapter, or slab-network-intelligence / slab-network-trace-cli for wireless traces.
Feedback -> Give the agent specific, observable results such as logs, debugger output, and expected vs actual behavior so it can iterate. See What to Expect.

What to Expect#

Describe a task, bug, question, or feature in natural language. The agent can then use AI SDK tools to search, inspect, generate, build, debug, and iterate.

Performance and LLM Dependency#

Result quality depends heavily on the LLM you use. The AI SDK provides relevant context and tools, but does not include the model itself.

More capable models generally produce better plans, tool usage, and implementation results.

Implementation Quality#

AI-generated code can contain bugs, just like human-written code. The value is faster iteration from your feedback and diagnostic output.

Serial monitors, debuggers, logs, and network tools give the agent stronger signals to work from.

Effective Feedback Strategies#

Scenario: You want to toggle an LED to indicate BLE connection, but it does not work.

Basic feedback

The onboard LED turns on when a BLE device is connected, but doesn't turn off when disconnected. Investigate and fix the issue.

More actionable feedback

I observe that the onboard LED turns on when a BLE device is connected, but doesn't turn off when disconnected. Add logging to critical parts of the application to help determine the issue. After that, rebuild and reflash the example, and monitor the serial output for 30 seconds. During this time, I will connect to and disconnect from the device. Use the logs to narrow down the possible causes.

Provide specific, actionable feedback and prompt the agent to use diagnostic tools such as serial monitor, GDB, and network intelligence for iterative troubleshooting.

Effective AI-Assisted Development#

Use a capable LLM for implementation, debugging, build, flash, and validation workflows.
State your goal clearly and include the project, board, SDK version, observed behavior, and success criteria when they matter.
Use silabs-knowledge-base when you need Silicon Labs APIs, components, examples, or compatibility guidance.
Give the agent specific positive or negative feedback from observable results, such as logs, debugger output, serial output, or expected versus actual behavior.

Topic	Description
Quick Reference	Common tasks and the related documentation, tools, and skills.
AI-Assisted Project	Create, modify, debug, and iterate on an AI-assisted example project.
Work with Existing Projects	Open and enhance an existing Silicon Labs project with AI assistance.
Overview	Product overview, features, and links to key topics.
Prompting Best Practices	Prompting guidance, model selection, MCP server selection, and validation practices.
Custom Hardware	Guidance for adapting Simplicity AI SDK workflows to custom hardware.
Installation Overview	Prerequisites, install options, and platform-specific guides.
OAuth Authentication Process	Configure OAuth authentication for Simplicity AI SDK.