digikey-mcp
A read-only MCP server for electronics engineers to search DigiKey parts, get detailed product information, plan and validate BOMs, and export order lists, using the official DigiKey API.
README
digikey-mcp
digikey-mcp is a read-only DigiKey Product Information V4 sourcing assistant for electronics
engineers. It provides candidate search, authoritative part details, requirements-led BOM
planning, refinement, validation, comparison, and fail-closed order-list export through a CLI and
a genuine MCP stdio server.
For AI agents: Read
AGENT.mdbefore installing, importing, configuring, or using this repository. It explains the MCP server, the companion skill, supported workflows, safety rules, and verification steps.
It does not place orders, change DigiKey accounts, check out, scrape the DigiKey website, or invent part numbers, prices, stock, specifications, URLs, or compatibility claims. Human review is required for critical electrical, thermal, EMC, and safety decisions.
Install
Python 3.11+ is required. uv is recommended for the reproducible clone-and-run workflow:
uv sync --all-groups
uv run python -m digikey_bom --help
The project exposes digikey-bom and digikey-mcp console scripts after installation.
uv is not required at runtime. To use the standard Python tooling instead:
python -m venv .venv
# macOS/Linux
source .venv/bin/activate
# Windows PowerShell: .venv\\Scripts\\Activate.ps1
python -m pip install --upgrade pip
python -m pip install .
digikey-mcp --help
For a published package, run python -m pip install digikey-mcp once the package is available
from your chosen package index. Offline mode works without DigiKey credentials or network access.
Configuration and modes
Copy .env.example to .env only for local configuration. Supported modes are:
offline(default): synthetic, visibly fixture-labeled data; no credentials or network.sandbox: DigiKey sandbox Product Information V4 API.production: DigiKey production Product Information V4 API.
Sandbox and production require a DigiKey developer subscription/application and the process
environment variables DIGIKEY_CLIENT_ID, DIGIKEY_CLIENT_SECRET, and DIGIKEY_ACCOUNT_ID for
ProductDetails calls. OAuth uses the two-legged client_credentials flow; tokens are cached in
memory only and never written to artifacts or logs. The client uses only official V4 keyword-search
and ProductDetails endpoints.
Keyword results are discovery evidence and may be stale. Current price and availability for a purchase-ready line come from ProductDetails. Rate limits and transient failures use bounded, injectable retries.
CLI
Offline examples work without credentials:
python -m digikey_bom search "24V to 5V 5A buck converter" --mode offline
python -m digikey_bom details DEMO-10K-0603-ND --mode offline
python -m digikey_bom bom "Build a Raspberry Pi 5 robot power and I/O BOM" --build-quantity 2 --format markdown --output-dir output
python -m digikey_bom validate-bom output/bom.json --mode offline
python -m digikey_bom export-bom output/bom.json --format tsv --mode offline
python -m digikey_bom refine-bom output/bom.json "must use 0603 package" --mode offline
python -m digikey_bom compare DEMO-10K-0603-ND DEMO-10K-REEL-ND --mode offline
digikey-mcp serve
Use --json on search, details, BOM, refine, validate, and compare for complete structured data.
export-bom emits the selected format to stdout. The default export excludes POSSIBLE_MATCH and
NO_VERIFIED_PART_FOUND; --include-possible-matches is explicitly unsafe review output.
MCP
The server exposes exactly these tools over stdio:
digikey_search_parts, digikey_get_part_details, digikey_build_electronics_bom,
digikey_refine_bom, digikey_validate_bom, digikey_export_order_list, and
digikey_compare_candidates.
Example Claude Desktop/Claude Code-style configuration:
{
"mcpServers": {
"digikey": {
"command": "uv",
"args": ["run", "--project", "/absolute/path/to/digikey-mcp", "digikey-mcp", "serve"],
"env": {"DIGIKEY_MODE": "offline"}
}
}
}
For live use, set DigiKey variables in the client environment. Do not paste secrets into prompts or skill instructions.
Evidence and artifacts
Every BOM line has one status:
VERIFIED_MATCH: exact DKPN observed in ProductDetails and every critical requirement verified.VERIFIED_WITH_ASSUMPTION: exact DKPN observed; remaining engineering assumptions are explicit.POSSIBLE_MATCH: exact DKPN observed, but a critical fact is missing, stale, or ambiguous.NO_VERIFIED_PART_FOUND: no safe part selected; failure and next action are retained.
The canonical artifact is schema 1.0.0, saved as output/bom.json by default. Optional views
include requested/purchase quantity, DKPN, MPN, package/MOQ, stock, price, lifecycle, datasheet,
product URL, and selection notes. Custom, user-provided, and non-DigiKey items remain separate.
The portable Claude-compatible skill is skills/digikey-bom/SKILL.md,
with workflow references and examples for robot power, CAN, sensor boards, multi-board BOMs, and
substitutions.
Agent handoff instructions are in AGENT.md.
Testing and development
Default tests are offline and use synthetic or mocked official-API-shaped responses. No real credentials or network calls are required:
uv run ruff format --check .
uv run ruff check src tests
uv run mypy src
uv run pytest -q
uv build
uv run twine check dist/*
See CONTRIBUTING.md and SECURITY.md.
Limitations
DigiKey price, stock, lifecycle, and availability data change. A returned part is not a complete electrical design, certification, thermal proof, safety case, EMC result, or schematic/layout review. Re-check ProductDetails before purchasing and obtain qualified human review for power, fault, battery, motor, E-stop, and other safety-critical decisions.
License
Copyright 2026 digikey-mcp contributors. Licensed under Apache-2.0; see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。