Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Description from the maker (original English)
OpenTable reservation management: find slots, book, cancel, list reservations, manage favorites.
These three are never merged into one score: opinions, measurements and popularity are different things.
We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.
Pick the AI app you use, then copy the settings.
OT_BRIDGE, OT_MCP_CHROME_URL, OT_WS_PORT, FETCHPROXY_WS_HOST, FETCHPROXY_IDENTITY_DIRYou need
Which app do you use?
Pick your app above. We remember it, so next time the button says “Install in …”.
Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json
{
"mcpServers": {
"opentable": {
"command": "npx",
"args": [
"-y",
"opentable-mcp"
],
"env": {
"OT_BRIDGE": "your-value",
"OT_MCP_CHROME_URL": "your-value",
"OT_WS_PORT": "your-value",
"FETCHPROXY_WS_HOST": "your-value",
"FETCHPROXY_IDENTITY_DIR": "your-value"
}
}
}
}Run this in your terminal
claude mcp add opentable --env OT_BRIDGE=your-value --env OT_MCP_CHROME_URL=your-value --env OT_WS_PORT=your-value --env FETCHPROXY_WS_HOST=your-value --env FETCHPROXY_IDENTITY_DIR=your-value -- npx -y opentable-mcpFill in the empty values before running it.
Paste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"opentable": {
"command": "npx",
"args": [
"-y",
"opentable-mcp"
],
"env": {
"OT_BRIDGE": "your-value",
"OT_MCP_CHROME_URL": "your-value",
"OT_WS_PORT": "your-value",
"FETCHPROXY_WS_HOST": "your-value",
"FETCHPROXY_IDENTITY_DIR": "your-value"
}
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
Paste into .vscode/mcp.json in your project
{
"servers": {
"opentable": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"opentable-mcp"
],
"env": {
"OT_BRIDGE": "your-value",
"OT_MCP_CHROME_URL": "your-value",
"OT_WS_PORT": "your-value",
"FETCHPROXY_WS_HOST": "your-value",
"FETCHPROXY_IDENTITY_DIR": "your-value"
}
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.
Your choice stays on this device in a cookie; no account needed.
OT_BRIDGE envOT_MCP_CHROME_URL envOT_WS_PORT envFETCHPROXY_WS_HOST envFETCHPROXY_IDENTITY_DIR envThese settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.
Runs on your computer; the tool list appears in your AI app after you install it.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 8, 2026
The registry name was verified with the GitHub account chrischall, which owns the repo
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
This one runs on your computer; there is no internet endpoint
Runs on your computer and uses local settings instead
Runs on your computer; we can only read tool lists from internet endpoints
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).
The GitHub repo last changed –, and the latest version (1.4.1) reached the registry 3 days ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.
In the same category: Spec Driven Development, Remote Desktop Commander, Funooy Krillion Reference
Servers in the same category from other makers, sorted by stars.
Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Funooy trivia references in six languages, with rules and fixed scoring. No daily spoilers.
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
OpenTable reservation manager as an MCP server for Claude — find slots, book, cancel, manage favorites, and read your dashboard via natural language.
v0.3.0-alpha status: Chrome-extension bridge, 10 tools, read + write. Every OpenTable request is relayed through your signed-in Chrome tab over a localhost WebSocket — each request rides your existing session and reaches OpenTable as if you'd clicked it yourself.
OpenTable's edge (Akamai Bot Manager) enforces a behavioral challenge on /, /s, /r/…, /dapi/…, and /booking/…. Tooling that builds its own HTTP client — cycletls, impersonated curl, headless Chrome — invents a separate identity and gets a 403 or JS interstitial. opentable-mcp does the opposite: it uses your own browser session as-is, with the cookies and TLS context it already has.
So instead of standing in for the browser, this MCP server:
127.0.0.1:37149 via @fetchproxy/server.fetch(..., { credentials: 'include' }) — your TLS, your cookies, your already-solved _abck./user/*) into tool-shaped output.No cookie-pasting. No cycletls. No Playwright. Just your own browser, acting on its own behalf — the MCP server only picks what to ask for.
| Tool | Kind | Source |
|---|---|---|
opentable_list_reservations | read | /user/dining-dashboard SSR |
opentable_get_profile | read | /user/dining-dashboard SSR |
opentable_list_favorites | read | /user/favorites SSR |
opentable_search_restaurants | read | /dapi/fe/gql?opname=Autocomplete |
opentable_get_restaurant | read | /r/{slug} SSR (__INITIAL_STATE__) |
opentable_get_menu | read | Restaurant profile SSR (restaurantProfile.menus) |
opentable_find_slots | read | /dapi/fe/gql?opname=RestaurantsAvailability |
opentable_book_preview | read | /booking/details SSR + SlotLock |
opentable_book | write | SlotLock → /dapi/booking/make-reservation |
opentable_cancel | write | /dapi/fe/gql?opname=CancelReservation |
opentable_modify_preview | read | /booking/details SSR (isModify=true) + SlotLock |
opentable_modify | write | /dapi/booking/make-reservation (isModify: true) |
opentable_add_favorite | write | /dapi/wishlist/add |
opentable_remove_favorite | write | /dapi/wishlist/remove |
opentable_healthcheck | read | /robots.txt (bridge probe) |
opentable_get_menu accepts the same restaurant ID, slug, path, or exact URL as opentable_get_restaurant. Omit menu_name to return all published menus, or pass an exact title such as "Dinner" (case-insensitive) to select one:
{"restaurant_id":"https://www.opentable.com/r/social-san-juan","menu_name":"Dinner"}
The response includes restaurant identity and source url, available_menus, menus, and an optional external menu_url. Each published menu retains OpenTable's sections/items, price strings, variationGroups, currency, provider, and updated timestamp. view defaults to compact (media URLs removed); full retains media URLs too. Published prices may differ from the restaurant's current menu, so keep the source and update timestamp when quoting.
Large menus come back as pages of whole items, so a reply never stops halfway through a dish. Without paging arguments a menu that fits is returned as-is; a larger one returns its first page. Each page carries available_sections and pagination:
{"offset":0,"returned_items":20,"total_items":57,"next_offset":20}
To continue, call again with offset set to next_offset and the same menu_name/section_name; next_offset is null on the last page. Optional paging inputs:
section_name — an exact section title (case-insensitive); the reply adds section_found.offset — the item to start from (default 0).limit — at most this many items per page (1–50, default 20). A page may return fewer to stay within the response budget.view: "full" with no paging inputs returns the whole menu uncapped, for clients without an output limit. A single item too large for a page is an error, never a truncated description.
status distinguishes available, menu_not_found (check available_menus), external_only, and not_available. An absent menu is not an HTTP/bridge or parsing error: those failures remain errors. External links are returned without fetching other hosts or parsing PDFs. This tool only reads the same restaurant page through the existing browser bridge; it never holds a table.
By using this MCP server, you acknowledge and agree to the following:
1. This server accesses your own OpenTable account. Every request is dispatched through your own signed-in browser tab via the ContextMint Bridge browser extension (or hangwin/mcp-chrome). It does not — and cannot — access anyone else's reservations.
2. [OpenTable's Terms of Use](https://www.opentable.com/c/legal/terms-and-conditions/) govern your use of this server, just as they govern your direct use of opentable.com. The clauses most relevant here:
You may not use any deep-link, robot, spider, scraper, generative AI or other AI technology including, but not limited to, those that operate by interacting with or otherwise making use of your browser, such as automated assistants or other automatic or manual device, process, or means to access, copy, search, or monitor any portion of the Services or OpenTable Content, except as expressly authorized by OpenTable.
And, critically: *"Actions of AI agents are acknowledged as actions of the User that is using them, and the User is responsible for checking and verifying the action…"*
You are agreeing to those terms — read by the maintainer 2026-05-23 — every time you invoke a tool in this server. OpenTable's ToU explicitly enumerates AI/automated-assistant access among the things they have not authorized; they also explicitly impute AI-agent actions back to you.
3. Personal, non-commercial use only. This project is not affiliated with, endorsed by, sponsored by, or in partnership with OpenTable, Inc. It is a personal automation tool that drives the same /dapi/... and /booking/... endpoints opentable.com uses. Do not use it to mass-book, sweep CC-required tables, resell reservations, or for any commercial purpose. The book/cancel/modify tools exist so you can manage one reservation as if you were sitting at your browser.
4. Stability is not guaranteed. This server depends on internal OpenTable persisted-GraphQL query hashes that OpenTable rotates between deployments. When they rotate, tools break and we re-capture them. Your own use is at the mercy of OpenTable's release cadence.
5. You accept full responsibility for any consequences of using this server in connection with your OpenTable account — rate limiting, slot-lock rejections, no-show penalties for runaway bookings, account warnings, suspension, or any enforcement action. Per OpenTable's ToU, the AI agent's actions are your actions — review every book/cancel before you confirm. If OpenTable objects to your use, stop using this server.
This section is the maintainer's good-faith summary of the terms — it is not legal advice and does not modify or supersede OpenTable's actual ToU.
npm install
npm run build
opentable-mcp shares one browser extension, ContextMint Bridge, with every other fetchproxy-based MCP. Install it once from https://github.com/nullnet-app/contextmint-bridge/releases:
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).
chrome://extensions → Developer mode → Load unpacked). Safari isn't available yet — it will ship inside the ContextMint app, which has no public download — so use Chrome for now.https://www.opentable.com/ in that same browser profile.After that, any MCP client that launches node dist/bundle.js will reach OpenTable through your signed-in tab.
Full setup + troubleshooting guide: see the ContextMint Bridge repo for the status-dot reference, WS protocol, and request lifecycle. Persisted-query hash capture for OpenTable redeploys is documented in CLAUDE.md here.
{
"mcpServers": { "opentable": { "command": "node", "args": ["/absolute/path/to/opentable-mcp/dist/bundle.js"] } } }
No env vars required by default — auth lives in the browser, not the MCP process.
If you've installed hangwin/mcp-chrome for browser automation, opentable-mcp can route its OpenTable fetches through it instead of ContextMint Bridge:
{
"mcpServers": { "opentable": { "command": "node", "args": ["/absolute/path/to/opentable-mcp/dist/bundle.js"], "env": { "OT_BRIDGE": "mcp-chrome" } } } }
In that mode you don't need ContextMint Bridge. Every OpenTable request becomes a chrome_network_request call against your existing mcp-chrome install, pinned via tabUrl to an opentable.com tab.
This mode is read-only. Booking, modifying, cancelling and favorites are refused with a clear error — OpenTable's write endpoints need a CSRF token from a restaurant/booking/account page, which this transport can't supply — and opentable_find_slots is unavailable. Use the default ContextMint Bridge transport for those.
Note: this path requires mcp-chrome ≥ the release containing PR #348 (tabUrl parameter on chrome_network_request). Pre-#348 mcp-chrome versions are active-tab-only and will misbehave for cross-origin fetches. Live-verification of this path is pending the upstream merge.
Other env vars: OT_WS_PORT (default 37149) overrides the fetchproxy WebSocket port; OT_MCP_CHROME_URL (default http://127.0.0.1:12306/mcp) overrides the mcp-chrome endpoint.
opentable_book, opentable_modify and opentable_cancel ask you before they act. A client that can show a confirmation prompt (Claude Code) shows one. On a client that cannot (claude.ai, Claude Desktop), the first call books, changes or cancels nothing: it returns a preview of exactly what would happen plus a confirmToken. Either way the prompt names the restaurant, date, time and party size, and for a booking or change the card that will be held and the cancellation policy (carried in the booking_token / modify_token from the preview tools; a cancel looks the reservation up on yo
Only the beginning is shown; read the rest in the repo