Embedded & Electronics Toolkit
← Back to EmbedCalc

Use EmbedCalc from Your AI

MCP server · embedcalc-mcp on npm · MIT

LLMs are confidently wrong about exactly the math embedded engineers care about: CRC check values, CAN bit-timing registers, struct padding, fixed-point quantization. Ask an AI for a CRC-16/MODBUS of a buffer and it will happily invent a plausible-looking wrong answer.

EmbedCalc MCP fixes this: it exposes EmbedCalc's calculators and code generators as Model Context Protocol tools, so Claude, Cursor, and any MCP-enabled agent call the exact, deterministic implementation instead of guessing — 29 tools, one command, no account.

Install

Claude Code:

claude mcp add embedcalc -- npx -y embedcalc-mcp

Claude Desktop / Cursor / any MCP client (JSON config):

{
  "mcpServers": {
    "embedcalc": { "command": "npx", "args": ["-y", "embedcalc-mcp"] }
  }
}

What your AI can now do exactly

Ask your AI…Tool it calls
"Give me MODBUS CRC-16 in C"crc_to_c → compilable table-driven C, check value 0x4B37
"CAN 500 kbit/s on a 36 MHz STM32F103"can_bit_timing → CAN_BTR = 0x00050008, SP 87.5%
"How big is this struct on a Cortex-M?"struct_layout → offsets, padding, total size
"0.5 in Q15?"qformat_convert → 16384 = 0x4000, quantization error
"Trace width for 3 A on 1 oz copper?"pcb_trace_width → IPC-2221 curve, mils & mm
"Nearest E24 to 4.67 kΩ?"eseries_nearest → 4.7 kΩ (+0.64%)
"Find me a real 220 Ω 1% resistor"find_component → in-stock MPNs + buy links

All 29 tools

Why trust the numbers

Every tool is ported from the same formulas that power the calculators on this site, and locked down with test vectors: standard CRC check values, oracle fixtures against the live CAN solver, IEEE-754 reference constants, ASTM/IPC published values. 99 tests run before every release. Source is on npm (embedcalc-mcp), MIT licensed.

Prefer clicking?

Every one of these tools is also a free interactive calculator on the EmbedCalc home page — same math, no AI required.