stm32-mcp-docs
MCP server that provides semantic search and documentation for STM32 microcontrollers, with specialized agents and slash commands for embedded development assistance.
README
STM32 MCP Documentation Server
An MCP (Model Context Protocol) server that provides semantic search over STM32 microcontroller documentation for use with Claude Code and other AI assistants. Features intelligent document chunking, ChromaDB vector storage, and 16 specialized agents for different embedded development domains.
Features
- One-Command Install: Install as a Claude Code plugin with one command
- Auto-Setup: MCP server, agents, and commands auto-configured on install
- Semantic Search: Find relevant documentation using natural language queries
- Peripheral-Specific Search: Filter results by STM32 peripheral (GPIO, UART, SPI, etc.)
- Code Examples: Retrieve working code examples for any topic
- HAL Function Lookup: Get documentation for specific STM32 HAL/LL library functions
- 16 Specialized Agents: Domain-specific agents for firmware, debugging, power, security, and more
- 4 Slash Commands: Quick access via
/stm32,/stm32-hal,/stm32-init,/stm32-debug - No Hardcoded Paths: Fully portable installation
Quick Start
Plugin Installation (Recommended)
# Install the STM32 plugin
/plugin install github:creativec09/stm32
This single command installs:
- MCP Server: Auto-configured via
mcp-config.json - 16 Agents: Available immediately for specialized STM32 assistance
- 4 Slash Commands:
/stm32,/stm32-hal,/stm32-init,/stm32-debug
Alternative: Manual MCP Installation
If you prefer not to use the plugin system, you can install just the MCP server.
Ubuntu 24 / WSL / Linux (auto-installs uv if needed):
claude mcp add-json stm32-docs --scope user '{"command":"bash","args":["-c","export PATH=\"$HOME/.local/bin:$PATH\" && (command -v uvx >/dev/null 2>&1 || curl -LsSf https://astral.sh/uv/install.sh | sh -s -- -q) && uvx --from git+https://github.com/creativec09/stm32.git stm32-mcp-docs"]}'
If you already have uv installed:
claude mcp add stm32-docs --scope user -- uvx --from git+https://github.com/creativec09/stm32.git stm32-mcp-docs
Note: For private repositories, include a GitHub Personal Access Token with repo scope in the URL: git+https://TOKEN@github.com/...
Start Using
After installation, restart Claude Code and use slash commands:
/stm32 How do I configure UART with DMA?
/stm32-init SPI master mode at 10MHz
/stm32-hal HAL_GPIO_Init parameters
/stm32-debug UART not receiving data
Or ask naturally:
"Show me how to configure GPIO interrupts on STM32H7"
"Why is my I2C peripheral returning HAL_TIMEOUT?"
Table of Contents
- Quick Start
- Installation
- MCP Tools
- Specialized Agents
- Usage
- Configuration
- Development
- Troubleshooting
- Uninstall
- License
Installation
Prerequisites
- Python 3.11 or higher (pre-installed on Ubuntu 24)
- curl (pre-installed on Ubuntu 24)
- Claude Code CLI
Note: uv is automatically installed on first use if not present.
Method 1: Plugin Installation (Recommended)
/plugin install github:creativec09/stm32
This installs everything automatically:
- MCP server configuration
- 16 specialized agents
- 4 slash commands
Method 2: Manual MCP + pip Installation
# Install the package
pip install git+https://github.com/creativec09/stm32.git
# Register with Claude Code
claude mcp add stm32-docs --scope user -- python -m mcp_server
Method 3: Development Installation
# Clone the repository
git clone https://github.com/creativec09/stm32.git
cd stm32-agents
# Create virtual environment
python -m venv .venv
source .venv/bin/activate # Linux/macOS
# .venv\Scripts\activate # Windows
# Install in development mode
pip install -e ".[dev]"
# Register with Claude Code
claude mcp add stm32-docs --scope user -- python -m mcp_server
MCP Tools
The server provides 15+ tools for STM32 documentation:
| Tool | Description |
|---|---|
search_stm32_docs |
Semantic search across all documentation |
get_peripheral_docs |
Get documentation for a specific peripheral |
get_code_examples |
Find code examples for a topic |
get_register_info |
Get detailed register documentation |
lookup_hal_function |
Look up HAL/LL function documentation |
troubleshoot_error |
Find solutions to errors and issues |
get_init_sequence |
Get peripheral initialization code |
get_clock_config |
Clock tree configuration examples |
compare_peripheral_options |
Compare peripheral features |
get_migration_guide |
Migration guides between STM32 families |
get_interrupt_code |
Interrupt handling examples |
get_dma_code |
DMA configuration examples |
get_low_power_code |
Low power mode configuration |
get_callback_code |
HAL callback examples |
get_init_template |
Complete initialization templates |
list_peripherals |
List available peripherals |
MCP Resources
| Resource URI | Description |
|---|---|
stm32://status |
Server status and statistics |
stm32://health |
Health check |
stm32://peripherals |
List documented peripherals |
stm32://stats |
Database statistics |
Specialized Agents
16 domain-specific agents are included in the plugin:
| Agent | Domain | Key Topics |
|---|---|---|
router |
Triage | Query classification, routing |
triage |
Triage | Initial query analysis |
firmware |
Core Development | General firmware questions |
firmware-core |
Core Development | HAL/LL, timers, DMA, interrupts, NVIC, RCC |
debug |
Debugging | HardFault analysis, SWD, trace |
bootloader |
Updates | Bootloader development |
bootloader-programming |
Updates | IAP, DFU, system bootloader |
peripheral-comm |
Communication | UART, SPI, I2C, CAN, USB, Ethernet |
peripheral-analog |
Analog | ADC, DAC, OPAMP, comparators, sensors |
peripheral-graphics |
Display | LTDC, DMA2D, DCMI, TouchGFX |
power |
Power | General power optimization |
power-management |
Power | Sleep, Stop, Standby, battery |
safety |
Certification | Safety-critical development |
safety-certification |
Certification | IEC 61508, ISO 26262, Class B |
security |
Security | Secure boot, TrustZone, crypto, RNG |
hardware-design |
PCB/Hardware | EMC, thermal, oscillators, layout |
See docs/AGENT_QUICK_REFERENCE.md for detailed agent capabilities.
Usage
Slash Commands
/stm32 <query> - General STM32 documentation search
/stm32-init <peripheral> - Get initialization code for a peripheral
/stm32-hal <function> - Look up HAL function documentation
/stm32-debug <issue> - Troubleshoot an STM32 issue
Natural Language Queries
Agents automatically search documentation:
"Show me how to configure GPIO interrupts on STM32H7"
"Why is my I2C peripheral returning HAL_TIMEOUT?"
"How to enter Stop mode and wake up on UART?"
Network Mode (Tailscale)
For accessing from multiple machines:
# Start server in network mode
STM32_SERVER_MODE=network python -m mcp_server --port 8765
# On client machines
claude mcp add stm32-docs --scope user --type sse --url "http://YOUR_TAILSCALE_IP:8765/sse"
Configuration
Configuration via environment variables:
STM32_SERVER_MODE=local # local, network
STM32_HOST=0.0.0.0 # Host to bind (network mode)
STM32_PORT=8765 # Port (network mode)
STM32_COLLECTION_NAME=stm32_docs # ChromaDB collection name
STM32_EMBEDDING_MODEL=all-MiniLM-L6-v2
STM32_LOG_LEVEL=INFO
Project Structure
stm32-agents/
├── .claude-plugin/ # Plugin manifest
│ └── plugin.json # Plugin configuration
├── agents/ # Agent definitions (16 agents)
├── commands/ # Slash command definitions
├── mcp-config.json # MCP server configuration
├── mcp_server/ # MCP server implementation
│ ├── server.py # Main server with tools/resources
│ ├── __main__.py # Module entry point
│ ├── config.py # Configuration management
│ ├── markdowns/ # Bundled STM32 documentation (80 files)
│ └── agents/ # Bundled agent definitions
├── pipeline/ # Document processing pipeline
├── storage/ # Vector storage layer
├── scripts/ # CLI utilities
├── tests/ # Test suite
└── docs/ # Comprehensive documentation
Development
Setup Development Environment
git clone https://github.com/creativec09/stm32.git
cd stm32-agents
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
Running Tests
pytest tests/
pytest tests/ --cov=mcp_server --cov=pipeline --cov=storage
Code Quality
black .
isort .
ruff check .
mypy mcp_server pipeline storage
System Requirements
| Requirement | Minimum | Recommended |
|---|---|---|
| Python | 3.11 | 3.12 |
| RAM | 4GB | 8GB |
| Disk Space | 1GB | 2GB |
| OS | Linux, macOS, Windows (WSL2) | Linux |
Performance Characteristics
| Metric | Value |
|---|---|
| First Run | 5-10 minutes (builds database) |
| Subsequent Starts | <5 seconds |
| Query Response | <100ms (warm) |
| Database Size | ~500MB |
| Total Chunks | 13,815 |
| Documentation Files | 80 |
Troubleshooting
"No documentation found"
Database may not have built yet. Wait for auto-ingestion on first run, or:
python scripts/ingest_docs.py --clear
Slow first request
First request takes 5-10 minutes due to:
- Loading embedding model
- Building vector database from 80 documents
Subsequent requests are fast (<100ms).
Server won't start
# Verify registration
claude mcp list
# Check server status
claude mcp status stm32-docs
# If uv issues, reinstall manually:
curl -LsSf https://astral.sh/uv/install.sh | sh
For more troubleshooting help, see docs/GETTING_STARTED.md.
Uninstall
Using Plugin System (Recommended)
If you installed via plugin:
# Remove the plugin (removes agents, commands, MCP config)
/plugin uninstall stm32-agents
# Clean up database (optional)
stm32-uninstall
Manual Uninstall
If you installed manually:
# Step 1: Remove MCP server configuration
claude mcp remove stm32-docs --scope user
# Step 2: Clean up database
stm32-uninstall
Uninstall Options
stm32-uninstall --dry-run # Preview what will be removed
stm32-uninstall --yes # Skip confirmation prompt
License
This project is licensed under the MIT License - see the LICENSE file for details.
Acknowledgments
Built with:
- MCP (Model Context Protocol) - AI integration framework by Anthropic
- ChromaDB - Vector database
- Sentence Transformers - Embedding models
- FastMCP - MCP server framework
Support
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.