Figma Context MCP
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
作者提供的说明(英文原文)
Unofficial AllTrails MCP for Claude — search trails, get details, reviews, photos, saved lists
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
ALLTRAILS_USER_ID, ALLTRAILS_REQUEST_TIMEOUT_MS, ALLTRAILS_DEBUG_LOG, ALLTRAILS_WS_PORT, ALLTRAILS_LOCALE, ALLTRAILS_CALLER你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"alltrails": {
"command": "npx",
"args": [
"-y",
"alltrails-mcp"
],
"env": {
"ALLTRAILS_USER_ID": "your-value",
"ALLTRAILS_REQUEST_TIMEOUT_MS": "your-value",
"ALLTRAILS_DEBUG_LOG": "your-value",
"ALLTRAILS_WS_PORT": "your-value",
"ALLTRAILS_LOCALE": "your-value",
"ALLTRAILS_CALLER": "your-value"
}
}
}
}在终端中运行这条命令
claude mcp add alltrails --env ALLTRAILS_USER_ID=your-value --env ALLTRAILS_REQUEST_TIMEOUT_MS=your-value --env ALLTRAILS_DEBUG_LOG=your-value --env ALLTRAILS_WS_PORT=your-value --env ALLTRAILS_LOCALE=your-value --env ALLTRAILS_CALLER=your-value -- npx -y alltrails-mcp运行前请先填写空着的值。
粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"alltrails": {
"command": "npx",
"args": [
"-y",
"alltrails-mcp"
],
"env": {
"ALLTRAILS_USER_ID": "your-value",
"ALLTRAILS_REQUEST_TIMEOUT_MS": "your-value",
"ALLTRAILS_DEBUG_LOG": "your-value",
"ALLTRAILS_WS_PORT": "your-value",
"ALLTRAILS_LOCALE": "your-value",
"ALLTRAILS_CALLER": "your-value"
}
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"alltrails": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"alltrails-mcp"
],
"env": {
"ALLTRAILS_USER_ID": "your-value",
"ALLTRAILS_REQUEST_TIMEOUT_MS": "your-value",
"ALLTRAILS_DEBUG_LOG": "your-value",
"ALLTRAILS_WS_PORT": "your-value",
"ALLTRAILS_LOCALE": "your-value",
"ALLTRAILS_CALLER": "your-value"
}
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
ALLTRAILS_USER_ID envALLTRAILS_REQUEST_TIMEOUT_MS envALLTRAILS_DEBUG_LOG envALLTRAILS_WS_PORT envALLTRAILS_LOCALE envALLTRAILS_CALLER env这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月8日
注册名称已通过 GitHub 账户 chrischall 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(3.2.1)于4天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:Figma Context MCP, Claude Real Video, Figma MCP Server
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Let any LLM watch a video locally — and search everything it has ever watched.
The Figma MCP server brings Figma design context directly into your AI workflow.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
A Model Context Protocol server that connects Claude to AllTrails, giving you natural-language access to trail search, trail details, reviews, photos, weather, and your own saved lists and hiking history.
[!WARNING]
AI-developed project. This codebase was entirely built and is actively maintained by Claude. No human has audited the implementation. Review all code and tool permissions before use.
[!CAUTION]
Unofficial and unsupported. AllTrails has no public API. This server reverse-engineers the internal API used by alltrails.com. That API is fronted by DataDome bot protection and governed by AllTrails' Terms of Service. It may break without notice, and automated access may violate their ToS — a prior third-party AllTrails project was disabled at AllTrails' request. Use this only for your own account, at your own discretion.
Ask Claude things like:
By using this MCP server, you acknowledge and agree to the following:
1. This server accesses AllTrails using your own signed-in session. It reuses the cookies from your own browser session; it cannot access anyone else's account or private data.
2. [AllTrails' Terms of Service](https://www.alltrails.com/terms) govern your use of this server, just as they govern your direct use of AllTrails. AllTrails restricts automated access and scraping, and fronts its API with bot protection. Automated use through this server may breach those terms. You are agreeing to AllTrails' terms every time you invoke a tool.
3. Personal use only. This project is not affiliated with, endorsed by, or sponsored by AllTrails, LLC. It is a personal automation tool. Do not use it to bulk-extract AllTrails' data, to build a competing product, or in any way that burdens their service.
4. You accept full responsibility for any consequences — technical (rate-limiting, CAPTCHA challenges, account action) or otherwise. The maintainer provides no warranty and no support.
This section is the maintainer's good-faith summary — it is not legal advice and does not modify or supersede AllTrails' actual ToS.
git clone https://github.com/chrischall/alltrails-mcp.git
cd alltrails-mcp npm install npm run build
Edit your Claude Desktop config file:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.jsonAdd the alltrails entry inside "mcpServers" (create the key if it doesn't exist):
{
"mcpServers": { "alltrails": { "command": "node", "args": ["/absolute/path/to/alltrails-mcp/dist/index.js"] } } }
Replace /absolute/path/to/alltrails-mcp with the path where you cloned the repo.
Quit completely (Cmd+Q on Mac, not just close the window) and relaunch.
Ask Claude: *"Search AllTrails for trails near me."*
AllTrails fronts its internal API with DataDome bot protection, and DataDome fingerprints the HTTP client itself — a cookie copied out of the browser and replayed from Node gets rejected even while the browser sails through. So the fetchproxy bridge is required: every API request runs as a same-origin fetch inside your own signed-in alltrails.com tab, reusing your authenticated session. There is no stored-cookie mode.
Requirements: the ContextMint Bridge browser extension installed (Chrome: download the chrome zip from the releases page and load it unpacked — use Chrome for now; Safari support will ship inside the ContextMint app, which has no public download yet), a signed-in alltrails.com tab open, and a one-time pair-code approval on first use (the trust persists).
ContextMint Bridge is the fetchproxy browser extension under its new name, from the same maintainer — fetchproxy's own README (https://github.com/chrischall/fetchproxy#extension) points to it. Its source is public at https://github.com/nullnet-app/contextmint-bridge: build it yourself, or check a release zip against the .sha256 file published beside it (shasum -a 256 -c contextmint-bridge-chrome- .zip.sha256).
The x-at-key app key AllTrails' own client sends is never stored in this repo or your config — the server captures the live value from your tab's own API traffic on first need, keeps it in memory only, and re-captures automatically if AllTrails rotates it. The alltrails_healthcheck tool round-trips a probe through the bridge and tells you which hop broke.
| Env var | Required | Purpose |
|---|---|---|
ALLTRAILS_USER_ID | No | Numeric user id for the per-user tools; defaults to the signed-in user via /api/alltrails/me. |
ALLTRAILS_WS_PORT | No | fetchproxy concentrator port (default 37149, shared by the whole fetchproxy fleet — override only for local dev/tests). |
ALLTRAILS_LOCALE / ALLTRAILS_CALLER | No | Override the corresponding request headers. |
ALLTRAILS_REQUEST_TIMEOUT_MS | No | Per-request timeout in ms (default 30000), applied on both paths. |
ALLTRAILS_DEBUG_LOG | No | 1/true/yes/on logs each request/response to stderr — method, path, status, response size, and a redacted 200-char request-body prefix (response bodies are never logged). |
All tools are read-only — this server never writes to AllTrails.
| Tool | What it does |
|---|---|
alltrails_search | Search AllTrails by trail or place name — free-text queries use the same suggestions endpoint as the alltrails.com search box (types: ["trail"] narrows to trails only) |
alltrails_resolve_location | Resolve a place name to AllTrails location records (country/state/city/area/POI) with each one's kind, coordinates, slug, and disambiguation label |
alltrails_get_trail | Trail details — name, overview, length, elevation gain, difficulty, rating, route type, location (detail: basic/medium/offline picks what AllTrails sends us) |
alltrails_get_trail_reviews | User reviews for a trail — {user, rating, comment} each |
alltrails_get_trail_photos | Photos for a trail — slim records with a fetchable image url |
alltrails_get_trail_gpx | Export a trail's route as GPX 1.1 (track points + elevation) |
alltrails_get_trail_weather | Weather overview for a trail |
alltrails_get_profile | The signed-in user's profile |
alltrails_list_user_lists | A user's saved lists / favorites |
alltrails_get_list_items | The trails saved in a list, by list id — ordered {trailId, type, order, notes, addedAt}, hydrate each via get_trail |
alltrails_list_completed_trails | A user's completed trails |
alltrails_get_activity_feed | A user's activity feed (feed: local/timeline/personal for the items, omit for the directory; cursor paginates) |
alltrails_healthcheck | Round-trips a probe through the fetchproxy bridge and reports role/port/timing plus a plain-English hint about which hop broke |
The per-user tools default to the signed-in user (resolved via /api/alltrails/me or ALLTRAILS_USER_ID); pass userId to target a public profile.
viewalltrails_search, alltrails_get_trail, alltrails_get_trail_reviews, alltrails_get_trail_photos, alltrails_get_list_items and alltrails_get_activity_feed take a view parameter with two rungs (the feed has a third, brief):
view | What you get |
|---|---|
compact | The default. The slim projection described in the table above — the fields a caller acts on. Nothing to pass. |
full | AllTrails' whole untouched record. |
brief | alltrails_get_activity_feed only: one line per item, who: what — — , reading the timeline feed when none is named. The context-groups (mcp-host) canonical *what's new* read. |
Compact is the default deliberately. These projections existed before, behind compact: false and a tool description asking the caller to please opt in — and an efficiency that has to be requested is one that usually is not. Ask for view: "full" when you need a field the projection drops; the commonest is route geometry from alltrails_get_trail.
`view` is not `detail`.
alltrails_get_trailalso takesdetail
(
basic/medium/offline), and that one is a passthrough to AllTrails —
it decides what the API sends *us*.
viewdecides what this server hands
*you*. Route geometry needs both:
detail: "offline"to fetch it and
view: "full"to keep it. The collision is exactly why the fleet parameter is
named
view.
One projection is not a subset:
alltrails_get_trail_photos'urlis
derived by the compact rung (AllTrails' record carries no image URL), so
view: "full"does not contain it.
403 Forbidden — AllTrails' DataDome protection rejected the request. This usually means the tab isn't signed in (or DataDome is challenging it) — sign into alltrails.com in an open tab and retry.
"AllTrails bridge: …" — the bridge itself failed before reaching AllTrails (extension not running, pairing not approved, no tab). Run alltrails_healthcheck for a diagnosis, and check the ContextMint Bridge extension popup.
"AllTrails: capturing the x-at-key app key failed…" — the key is read off a request your tab makes while the call is waiting, and an idle tab makes none. Refreshing first does not help: the extension only listens for the duration of the call, so a reload that finished before it started happened while nothing was listening. Run the tool again and reload a signed-in www.alltrails.com tab while it is running.
Empty / unexpected results — the internal AllTrails endpoints are undocumented and change over time; responses may shift. Enable ALLTRAILS_DEBUG_LOG=1 to trace each request's method, path, status and response size on stderr.
npm test # run the vitest suite
npm run build # tsc → dist/, then esbuild bundle → dist/bundle.js npm run dev # node --env-file=.env dist/index.js (requires built dist)
vitest.config.ts enforces 100% coverage on src/** (except the stdio entry point). Main is protected — all changes land via PR. See CLAUDE.md for the full PR + release flow.
src/
index.ts MCP server entry (runMcp + StdioServerTransport) protocol.ts Wire-level constants (BASE_URL, app key, headers) transport.ts createAllTrailsTransport(): the fetchproxy bridge transport client.ts AllTrailsClient — bridge requests, live x-at-key capture, 429 retry config.ts Env parsing (api key, headers, timeout, port, user id, debug) valida
只显示了开头部分;其余内容请在仓库中阅读