✨ Open infrastructure for self-healing robots and software-defined machines.

selfpatch is an open, practical attempt to build the missing "nervous system" for robots and SDVs: diagnostics, introspection, and update flows that are good enough for humans and for AI agents to reason about.

This GitHub organization hosts the open-source core and experiments around that idea.

Modern robots and SDVs run on complex stacks:

• ROS 2 graphs, nodes, topics, actions
• Mixed hardware (MCUs, ECUs, edge HPC, cloud)
• Many protocols (DDS, Zenoh, WebSockets, Iceoryx, UDS, OPC UA, CAN/J1939, …)

Yet diagnostics and updates are often:

• Fragmented per vendor / per subsystem
• Hard to introspect at runtime
• Designed for humans clicking through tools, not for automation or AI

selfpatch aims to change that by providing:

1. A modern diagnostic & introspection layer

   • Runtime discovery of components, apps, topics, health, configuration
   • API-first approach: HTTP/REST + schemas instead of ad-hoc scripts

2. A foundation for safe self-healing flows

   • Clear, machine-readable model of the system
   • Hooks for health checks, mitigation actions, and OTA-style updates
   • Designed so that AI agents can understand and justify their actions

3. Bridges instead of rewrites

   • Integrations with existing ecosystems (ROS 2, UDS, SOVD, OPC UA, …)
   • Evolution over revolution: start where your robots are today.

Repository	Description	Status
ros2_medkit	SOVD-compatible REST gateway for ROS 2 diagnostics
sovd_web_ui	React web UI for browsing SOVD entity trees
ros2_medkit_mcp	Model Context Protocol server for AI agent integration	🚧 In Progress
selfpatch_demos	Demo integrations (TurtleBot3, Nav2)

Modern, SOVD-compatible diagnostics for ROS 2 robots — C++17 gateway exposing the ROS 2 graph via REST API.

Entity Model (SOVD-aligned):

Key Features:

• 🔍 Runtime Discovery — Automatically discovers running nodes, topics, services, actions
• 🌐 REST API — HTTP endpoints for all entity types, data, operations, configurations
• 📊 Dynamic Serialization — Read any ROS 2 message type at runtime (via dynmsg)
• 📡 Data Access — Read or publish data over ROS 2 topics via HTTP
• ⚙️ Configuration — Get/set ROS 2 parameters through REST
• 🎯 Operations — Invoke services and send action goals via HTTP
• ⚠️ Fault Management — Unified fault reporting with snapshots capturing system state at fault time
• 📄 Manifest Support — Hybrid discovery with YAML manifests for static entity definitions

# Quick start
ros2 launch ros2_medkit_gateway gateway.launch.py
curl http://localhost:8080/api/v1/areas

📖 Documentation • 💬 Discord

Lightweight React SPA for browsing SOVD entity trees — connects to ros2_medkit gateway and visualizes the diagnostic hierarchy.

Tech Stack: React 19 + TypeScript + Vite + TailwindCSS + shadcn/ui + Zustand

Features:

• 🌳 Entity Tree Browser — Hierarchical navigation with lazy-loading
• 📂 Virtual Folders — data/, operations/, configurations/ per entity
• 📊 Topic Viewer — Real-time data display with QoS information
• ⚙️ Parameter Editor — View and modify ROS 2 parameters
• 🎯 Operation Invoker — Call services and send action goals from the UI

# Quick start
docker run -p 8080:80 ghcr.io/selfpatch/sovd_web_ui
# Then connect to your ros2_medkit gateway URL

Model Context Protocol (MCP) server wrapping ros2_medkit REST API for LLM tool use.

🔒 Read-only by design — safe for AI agents to explore without risk of modifying the system.

Enables AI agents (Claude, GPT, etc.) to:

• 🔍 Discover and query robot components
• 📊 Read sensor data and system state
• 📋 List available configurations and operations
• 🔧 Diagnose and troubleshoot issues

# Start MCP server
ROS2_MEDKIT_BASE_URL=http://localhost:8080/api/v1 poetry run ros2-medkit-mcp-stdio

Available Tools:

• sovd_entities_list — Discover areas and components
• sovd_component_data — Read topic data
• sovd_list_operations — List available services/actions
• sovd_list_configurations — Get parameters
• sovd_faults_list — List active faults

Real-world demonstrations of ros2_medkit with ROS 2 robots.

Available Demos:

# Run the TurtleBot3 demo
cd demos/turtlebot3_integration
docker compose up
# Open http://localhost:8080 for Web UI

┌─────────────────┐                   ┌─────────────────┐
│  sovd_web_ui    │                   │ ros2_medkit_mcp │
│  (React SPA)    │                   │ (MCP for LLMs)  │
└────────┬────────┘                   └────────┬────────┘
         │ HTTP/REST                           │ HTTP/REST
         │ /api/v1/...                         │ /api/v1/...
         └──────────────┐     ┌────────────────┘
                        ▼     ▼
               ┌─────────────────────┐
               │ ros2_medkit_gateway │
               │  (C++ ROS 2 node)   │
               └──────────┬──────────┘
                          │ ROS 2 APIs
               ┌──────────▼──────────┐
               │    ROS 2 System     │
               │ (nodes, topics,     │
               │  services, actions, │
               │  parameters, faults)│
               └─────────────────────┘

• Robotics teams using ROS 2 who need better remote diagnostics and observability
• SDV / mobility engineers looking to modernize diagnostics without rewriting everything
• AI/ML engineers building autonomous diagnostics and remediation with LLMs
• Tooling & platform teams building monitoring, OTA, or AI-driven operations
• Researchers & tinkerers exploring self-healing, digital twins, and autonomous remediation

If you've ever thought "we can't safely automate fixes because we don't really understand what's running where", you're in the right place.

We're actively developing all repositories. You can help by:

• ⭐ Starring the repositories
• 🐛 Opening issues with use-cases, pain points, and ideas
• 💻 Contributing code, docs, or examples
• 🤖 Testing with your own robots and sharing feedback
• 💬 Joining discussions on Discord

See individual repository CONTRIBUTING.md files for guidelines.

• 💬 Discord — Join our server for discussions and support
• 📖 Documentation — selfpatch.github.io/ros2_medkit
• 🐛 Issues — Open issues in individual repositories
• 📢 Watch — Star and watch this organization for updates

If you're working on robots, SDV platforms, or tooling and this resonates, don't hesitate to reach out – collaboration is the whole point of this effort.