The Problem

Debugging firmware on nRF devices is notoriously tedious. You flash your firmware, open a terminal, and watch raw logs (RTT/UART) scroll past. Trying to correlate timestamps between two boards, decipher hex codes, and manually search your source code to find where an error originated is a major productivity bottleneck.

nRF AI Debugger changes that. It's an AI agent built specifically for the nRF Connect SDK ecosystem. It captures live logs directly from your boards and analyzes them in real-time, correlating firmware output with your source code to pinpoint the root cause of failures.

Features

📊 1. Capture & Analyze Device Logs

The nRF AI debugger captures live RTT or UART logs, identifies patterns in your application's behavior, and produces structured analysis reports—covering everything from boot sequences to protocol-specific events.

Key Capabilities:

• 🔍 Auto-detects connected boards via J-Link serial numbers.
• 📡 Multi-device capture — two devices (or more) simultaneously (e.g. Central + Peripheral).
• 🧠 Context-aware analysis — correlates logs with your actual main.c and project files.
• 💡 Proactive Debugging — catches Hard Faults or stack overflows and points to the offending line of code.

🔧 2. Generate Best-Practice Logging

Before you can analyze, you need good logs. The agent reads your nRF Connect SDK project, understands the BLE stack, and injects the right log statements — so when it analyzes later, it knows exactly what each line means.

Key Capabilities:

• 📁 Multi-project awareness — handles Central + Peripheral workspaces simultaneously
• ⚙️ Auto-configures logging backend (RTT vs UART) in prj.conf
• 🎯 NCS-compliant — follows Zephyr RTOS logging best practices
• 🔘 Interactive — asks before modifying, suggests RTT over UART for BLE projects

Why does the agent generate the logging code? Because an agent that wrote the log statements can analyze the output far more intelligently — it understands the context because it created it.

Quick Start

1. Install nRF AI Debugger from the VS Code Marketplace.
2. Configure your AI provider (We recommend GLM-4.7 for cost-effective, high-performance analysis).
3. Choose a mode: "Analyze nRF Device Logs" or "Generate Logging Code".

Requirements

Roadmap & Compatibility

We are expanding based on community needs. If you need support for a specific protocol or board, join our discussions!

Tested Models

"We optimized for these models so you can debug for hours for the price of a cup of coffee."

New to GLM? Follow the Step-by-Step Configuration Guide to get your API key and set up the OpenAI-compatible endpoint in VS Code.

🔒 Privacy & Security

Your firmware stays yours.

• Local Control: The agent runs entirely on your machine. It only sends specific log snippets and code context to your chosen AI provider.
• BYOK (Bring Your Own Key): You have full control over which model you use and which API endpoint you trust.
• Open Source: Our capture scripts and agent logic are fully transparent and auditable by the community.

How It Works

graph LR
    A[Your nRF Project] -->|Agent reads code| B[Log Generator]
    B -->|Injects log statements| A
    A -->|Build & Flash| C[nRF Device]
    C -->|Live RTT/UART capture| D[Log Analyzer]
    D -->|Code-aware analysis| E[Expert Report]

About Us

Adsum Networks — We've been developing IoT solutions on nRF and other embedded platforms for over 7 years. We built nRF AI Debugger because we needed it ourselves to handle complex BLE debugging, and now we're sharing it to help the community build better firmware, faster.

Acknowledgments

• Cline — The open-source AI assistant this project builds upon.
• Nordic Semiconductor — For the exceptional nRF Connect SDK and developer tools.

License

Apache 2.0 © 2026 Adsum Networks