"Open Collective" matching MCP servers:

Check 22,000+ devices with open firmware or open protocols (OpenWrt, LineageOS, coreboot, Zigbee2MQTT, Tasmota…) before reverse engineering one; claim a device while you work on it and publish your findings when done.

Reverse engineer Windows-only USB gadgets: capture vendor-app traffic, infer the protocol with frame-level evidence, and generate an open Python/pyusb driver. Replay needs human confirmation.

Enables MCP clients to list serial ports, open sessions, send firmware commands, read and filter output, and safely flash STM32 or ESP32 boards with explicit confirmation and allow-listed firmware paths.

MCP server for serial port communication. Provides tools to open, read, write, and manage serial ports through the Model Context Protocol.

ChannelTerm is an open-source MCP server that enables humans and AI agents to share live serial and SSH terminal sessions for embedded systems development and debugging. It allows AI agents to read device output, execute commands, assist with GDB debugging, and transfer files while keeping developers in control.

Let an AI agent drive up to 12 hobby servos on an open-hardware USB-C controller (Pololu Maestro compatible): moves, named poses, sequences, on-board scripts and rail current/voltage readout, with per-channel travel limits and overcurrent guardrails. Includes a --mock emulator, so it runs without the board.

Enables AI coding assistants to drive Xilinx Vivado through natural language, letting users open .xpr projects, run synthesis, implementation, and bitstream generation, inspect timing, resource, and clock reports, and launch simulations to read signal values or set breakpoints. It also exposes design hierarchy queries and arbitrary TCL execution, so FPGA design flows can be operated conversationally on Windows.

Enables MCP clients like Cursor to control logic analyzers, open capture files, decode buses such as Intel 8080, and reconstruct display framebuffers like ST7789.

An open-source MCP server that turns natural language hardware descriptions into complete, buildable projects including circuits, firmware, 3D-printable enclosures, PCB files, and documentation.

Let an AI agent sniff and reverse-engineer a car's LIN bus with an open-source USB LIN dongle: per-id capture with changing bits, change log, auto-detect speed/checksum, LDF decode/encode; read-only by default, master/slave simulation behind --allow-transmit, --mock simulated bus.

Gives AI assistants a fully offline toolchain for FlyThings OS human-machine interface development: converting HTML prototypes into JSON/FTU layouts, packaging and deploying firmware to devices, capturing real-device screenshots for pixel-diff acceptance, and generating i18n, tests, and bug reports. It also bundles a local vector-plus-BM25 knowledge base covering hardware models, UI control semantics, dependency manifests, and reusable components.

An open-source MCP server that bridges AI models with industrial equipment, supporting multiple protocols like Modbus, OPC UA, and MQTT for reading data and controlling machines.

Enables agents to query LAVA board farms, submit and manage test jobs, and open interactive board sessions via the LAVA REST API.

A fully local MCP server for Mixly that lets AI clients discover boards, scan blocks, build/validate/open projects, generate code, and compile via Arduino CLI—all without uploading source or relying on a remote server.