The Evolution Beyond Cortex-Debug

Introducing mcu-debug

A modern, modular embedded debugging ecosystem engineered for automation, headless CI workflows, seamless remote development, and AI agent integration.

CLI & CI First

Full TUI mode and headless command-line interface. Automate firmware flashes, GDB trace captures, and regression tests in CI pipelines.

Zero VS Code dependency for CI

Frictionless Remote

First-class support for WSL2, Docker containers, and SSH targets with zero manual port forwarding or pipe setup.

Native cross-platform tunnels

AI Agent Native

Multiplexes raw GDB commands, SVD registers, RTT buffers, and logs into a structured stream for hybrid AI/human debugging.

Raw GDB feature access

Drop-in migration path for Cortex-Debug users Open Source & Community Driven

Pillar 1 & 2

Automation & Remote Infrastructure

Bring your embedded debugging out of isolated desktop IDEs into modern cloud & containerized dev pipelines.

CLI & CI Automation

Headless Mode

Launch full debug sessions directly from the command line or within automated test suites without booting a graphical user interface.

TUI Debugger: Rich terminal user interface for interactive terminal-only environments.
CI Pipeline Ready: Run automated hardware-in-the-loop (HIL) tests in GitHub Actions or Jenkins.
Unified Scripting: Programmatically execute GDB commands and capture memory dumps.

$ mcu-debug debug --config "Launch CM0" --no-tui
$ mcu-debug debug --config 0

Frictionless Remote Debugging

Zero Config Tunnels

Native support for remote development environments without standard remote GDB networking gymnastics.

Docker & Container Dev: Seamless hardware passthrough into Linux dev containers.
WSL2 Native Speed: Direct USB debug probe access with low-latency sockets.
Remote SSH Hosts: Debug hardware attached to remote farm servers transparently.

"hostConfig": { "type": "auto" }
"hostConfig": true

Reduces setup time from hours of custom shell scripts to a single standardized configuration file.

Pillar 3

AI-Assisted Debugging & Hardware Telemetry

Multiplexing raw GDB capabilities, register states, and memory streams into a coherent AI diagnostic surface.

Human Developer (Strategic Intent)

VS Code / CLI TUI

Session Notes

High-Level Decisions

Structured Session Context & Guidance

 Autonomous AI
                                Debug Agent
 Direct GDB Interface
 Autonomous
                                    Inspection
 Cause Hypothesis
                                

Multiplexed Telemetry & Debug Stream

 Target
                                Hardware & Telemetry Engine
 GDB Control Engine
 RTT Memory Buffer
 SVD Peripherals
 Serial/UART Log
                                

[RTT#0] Task execution delayed in DMA queue (0x20001400)

[UART:Debug] USART1->ISR: TXE=0, TC=0, ORE=1 (Overrun Error)

[GDB] frame #0: HardFault_Handler () at faults.c:42

[AI-REQUEST] Isolated: RX overrun triggered unhandled ISR spinlock.

Multiplexed Debug Surface

RTT + SVD + GDB Commands + UART

Unifies fragmented hardware output channels into a clean, machine-parsable stream. AI agents don't struggle with raw terminal noise; they read structured hardware telemetry.

Autonomous Raw GDB Operations

Full Direct Control

The AI agent directly executes low-level GDB inspection commands—evaluating pointers, reading register states, dumping memory maps, and setting conditional hardware breakpoints autonomously.

Hybrid Mode & Session Notes

Human-in-the-Loop

Developers maintain full strategic direction. Using session-notes, the human sets targets while the AI logs hypotheses and asks for physical board actions (e.g., unplug cable, toggle jumper).

Provides LLM agents with real hardware visibility rather than guessing from static source code alone.

Strategic Transition

Drop-in Evolution from Cortex-Debug

Built on years of cortex-debug experience, upgraded for next-generation embedded demands.

Frictionless Team Onboarding

Because mcu-debug shares DNA with cortex-debug, existing launch configurations can be reused with minimal modifications.

Legacy launch.json

"type": "cortex-debug"

New mcu-debug launch.json

"type": "mcu-debug"

Key Strategic Advantages

Standardized Tooling: One debugger for local VS Code, remote Linux servers, and automated CI tests.
AI Readyness: Prepares team workflows for AI co-pilots with multiplexed telemetry feeds.
Vendor Neutrality: Works across ARM Cortex-*, RISC-V, OpenOCD, J-Link, ST-Link, probe-rs, and PyOCD. 64-bit ready

Available on GitHub & soon on VS Code Marketplace

Ready to explore mcu-debug?

Try running a test session side-by-side with your existing setup. Fully open source and built to empower modern embedded engineering teams.

https://github.com/mcu-debug/mcu-debug