Lets an AI assistant search a product directory, compare alternatives, and manage product listings via an MCP server.
Also known as makerradar-io/makerradar-cli, makerradar-cli
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 1.0.0
- Last commit
- Nov 12, 2026
- License
- MIT
What is MakerRadar MCP?
MakerRadar lets an AI assistant search a product directory, compare alternatives, and manage product listings via an MCP server.
Summary generated from the project's README
About MakerRadar MCP server
MakerRadar provides an MCP server that connects AI assistants to a product directory and launch platform. Developers and AI agents can use this to search for products, compare different software alternatives, and explore various categories or weekly launch boards. The server is available remotely via a public endpoint, meaning no local installation is required for basic directory searches.
When users need to manage their own product listings, they can connect an account using OAuth or an agent token. This allows an AI client to perform write actions such as drafting new listings, updating existing ones, or managing images and launch options. The system is designed for compatibility with clients like Claude Desktop, Cursor, and VS Code. It includes specialized tools for site analysis, such as checking SEO, security, and technical stacks, making it a versatile tool for product research and management.
What you can do with it
- 1Research product alternativesAsk the AI to find and compare different software options for a specific task.
- 2Draft new product listingsHave an agent create a preliminary listing for a new product website.
Written from the project's README. Every use case is checked against a quote from it.
How to install MakerRadar MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall MakerRadar MCP in Claude Code
$ claude mcp add --env 'MAKERRADAR_TOKEN=<YOUR_MAKERRADAR_TOKEN>' makerradar-cli -- npx -y makerradarInstall MakerRadar MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"makerradar-cli": {
"command": "npx",
"args": [
"-y",
"makerradar"
],
"env": {
"MAKERRADAR_TOKEN": "<YOUR_MAKERRADAR_TOKEN>"
}
}
}
}Install MakerRadar MCP in VS Code
.vscode/mcp.json
{
"servers": {
"makerradar-cli": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"makerradar"
],
"env": {
"MAKERRADAR_TOKEN": "<YOUR_MAKERRADAR_TOKEN>"
}
}
}
}Install MakerRadar MCP in VS Code CLI
$ code --add-mcp '{"name":"makerradar-cli","type":"stdio","command":"npx","args":["-y","makerradar"],"env":{"MAKERRADAR_TOKEN":"<YOUR_MAKERRADAR_TOKEN>"}}'Install MakerRadar MCP in Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"makerradar-cli": {
"command": "npx",
"args": [
"-y",
"makerradar"
],
"env": {
"MAKERRADAR_TOKEN": "<YOUR_MAKERRADAR_TOKEN>"
}
}
}
}Install MakerRadar MCP in Codex
$ codex mcp add makerradar-cli --env 'MAKERRADAR_TOKEN=<YOUR_MAKERRADAR_TOKEN>' -- npx -y makerradarInstall MakerRadar MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.makerradar-cli] command = "npx" args = ["-y", "makerradar"] [mcp_servers.makerradar-cli.env] MAKERRADAR_TOKEN = "<YOUR_MAKERRADAR_TOKEN>"
Install MakerRadar MCP in Gemini CLI
$ gemini mcp add -s user -e 'MAKERRADAR_TOKEN=<YOUR_MAKERRADAR_TOKEN>' makerradar-cli npx -- -y makerradarInstall MakerRadar MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"makerradar-cli": {
"command": "npx",
"args": [
"-y",
"makerradar"
],
"env": {
"MAKERRADAR_TOKEN": "<YOUR_MAKERRADAR_TOKEN>"
}
}
}
}Configuration
MAKERRADAR_TOKENSecretAn agent token (mr_…) from `makerradar login`, only with --account: manage your own listings.
| Name | Description | |
|---|---|---|
| MAKERRADAR_TOKEN | An agent token (mr_…) from `makerradar login`, only with --account: manage your own listings. | Secret |
Automated checks
Ran Nov 14, 2026Full report →Passed our automated checks. That is not a security audit: read the code and give the server the narrowest access it needs. How checks work
- Source repository linked and reachablemakerradar-io/makerradar-cli is public and reachablegithub.com/makerradar-io/makerradar-cli ↗Pass
- Package names this repository as its sourcemakerradar names makerradar-io/makerradar-cli as sourcenpmjs.com/package/makerradar ↗Pass
- Build provenance (Sigstore attestation)No build attestation published for this releaseNot checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-12github.com/makerradar-io/makerradar-cli ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNo popular package one edit away from makerradarPass
- No critical advisory in the package or its dependencies1 resolved dependencies checked; no critical advisoryosv.dev ↗Pass
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeinitialize answered with protocol 2025-06-18 in 230 msPass
README
View on GitHubMakerRadar CLI
The official command-line client for MakerRadar, the launch platform and product directory. Search the directory, follow the weekly launch board, draft a listing for your product and manage your listings from the terminal, by hand or from a script or an AI agent.
npx makerradar search "invoicing"
- No account needed to search, read products, see launches and draft a listing.
- Your account, safely:
makerradar loginconnects with a code you approve in the browser. The CLI never sees your password, can't vote, comment or review, and never pays. - Made for agents too:
--jsonprints the API's own JSON, errors are JSON on stderr, and exit codes are stable.
Docs: makerradar.io/cli · API reference: makerradar.io/developers · MCP server: makerradar.io/mcp
Install
Needs Node.js 20 or newer.
npm install -g makerradar
makerradar --help
Or run it without installing: npx makerradar <command>.
Commands
Directory (no account needed)
| Command | What it does |
|---|---|
makerradar search [query] |
Search listed products, newest first. Filters: --category, --platform, --audience, --use-case, --country, --updated-since; pages: --limit (≤ 100), --cursor |
makerradar product <slug|url> |
One product in full: website, pricing, upvotes, rating, tags, links |
makerradar alternatives <slug|url> |
Listed alternatives to a product, ranked as on its alternatives page |
makerradar categories |
Every category, with how many products it has |
makerradar launches [week] |
This week's launch board, or another week's: 45, 2026-W45 or week-45-2026 |
makerradar pricing |
What's on sale (Premium launch, featured weeks, ads…), with prices |
makerradar draft <website> |
Draft a listing for your product. You get a link to review and submit it: nothing is public until you do, and every listing is reviewed first |
Free site checks (no account needed)
| Command | What it does |
|---|---|
makerradar check <url> |
Check a page or site with one of MakerRadar's free tools: on-page SEO by default, any other with --tool <slug>. Every check comes back GOOD / IMPROVE / FIX with what to do, then whether the site is listed |
makerradar tools |
Every free tool and the command that runs it. check runs the website grader, SEO, SERP snippet preview, redirects, broken links, structured data, social card, sitemap, robots.txt, logo, AI visibility, AI citation readiness, what AI sees, llms.txt, agent readiness, security headers, email DNS, domain age, certificate, hosting, tech stack, alternatives, domain availability and Domain Rating (needs login) |
makerradar keywords <words> |
Search phrases people use to find a kind of product: buyer, comparison, how-to and questions people ask AI |
makerradar robots |
robots.txt rules for AI crawlers (GPTBot, ClaudeBot, PerplexityBot …): --preset allow-all|block-training|block-all, --block / --allow <token>, --disallow <path>, --sitemap <url>, --signals search,ai-input,ai-train|none. The file goes to stdout (> robots.txt), what it blocks to stderr |
Your account
| Command | What it does |
|---|---|
makerradar login |
Connect the CLI to your account: open the link, sign in, approve. --scope (repeatable) limits what it may do: listings:read, listings:write, launches:write, billing:checkout |
makerradar whoami |
The account the CLI acts for, and what it may do |
makerradar listings [id] |
Your listings, or one of them in full |
makerradar logout |
Forget the saved token on this machine (revoke it in Dashboard › AI agents) |
Agents and developers
| Command | What it does |
|---|---|
makerradar mcp [client] |
Connect an AI assistant to MakerRadar's MCP server: claude-code, claude-desktop, cursor, vscode, windsurf, stdio |
makerradar mcp serve [--account] |
Run the MCP server over stdio for a client that can only start a local command (see below) |
makerradar api <METHOD> <path> |
Call any public API endpoint and print its JSON. --data takes JSON, @file or - (stdin); your token is sent when you're logged in |
makerradar search crm --platform ios --limit 5
makerradar product https://makerradar.io/p/acme-analytics
makerradar launches 2026-W45 --json
makerradar draft https://acme.example --tagline "Invoices in one click"
makerradar api PATCH /me/listings/prd_123 --data '{"tagline":"Invoices in one click"}'
makerradar mcp claude-code
npx -y makerradar mcp serve # the MCP server over stdio
MCP server
MakerRadar's MCP server lets an AI assistant search the directory and, signed in, manage your
listings. It's remote — nothing to install — and this repository is its public home: the
registry entry (io.makerradar/mcp), the setup for every client, and a stdio bridge.
| Server | URL | Sign-in |
|---|---|---|
| Public | https://api.makerradar.io/mcp |
none — search, products, alternatives, comparisons, categories, launch weeks, draft_listing |
| Account | https://api.makerradar.io/mcp/account |
OAuth (a "Connect" button), or an agent token — your listings, launches, ownership proof, images, checkout |
Both speak Streamable HTTP (protocol 2025-06-18 and later); the server card is at
/.well-known/mcp/server-card.json and the
docs at makerradar.io/mcp. Every tool carries a title and a
read-only / write annotation. No tool votes, comments or reviews; a checkout hands the person a
payment page — the agent never pays.
Set it up in one line. makerradar mcp <client> prints the exact config for your client:
npx makerradar mcp claude-code # runs the two `claude mcp add` commands to copy
npx makerradar mcp cursor # ~/.cursor/mcp.json
npx makerradar mcp vscode # .vscode/mcp.json
npx makerradar mcp windsurf # ~/.codeium/windsurf/mcp_config.json
npx makerradar mcp claude-desktop # Settings › Connectors › Add custom connector
A client that can only start a local command (or a catalog that lists installable servers) runs the bridge instead. It relays each JSON-RPC message to the remote server and prints every answer — nothing is interpreted locally:
{
"mcpServers": {
"makerradar": { "command": "npx", "args": ["-y", "makerradar", "mcp", "serve"] }
}
}
Add "--account" to the arguments (and sign in with makerradar login, or set MAKERRADAR_TOKEN)
to serve /mcp/account with your token so the assistant can manage your listings.
Tools on the public server: search_products, get_product, get_alternatives,
compare_products, list_categories, list_current_launches, get_launch_week, top_products,
draft_listing, get_pricing, connect_account / connect_account_finish, and the free site
checks — check_page_seo, check_ai_visibility, check_site_security, check_email_dns,
lookup_domain, check_tls, check_hosting, detect_tech (any public website; the same checks as
https://makerradar.io/tools). On the account server,
also: get_my_account, list_my_listings, get_my_listing, submit_listing, update_listing,
get_listing_options, set_quick_answers, add_listing_image, remove_listing_photo,
set_listing_cover, get_launch_options, choose_launch, start_ownership_check,
verify_ownership, get_badge_snippet, start_checkout, list_my_orders.
As a plugin. The repository is also an Agent Plugin: plugin.json,
mcp.json (both servers) and the skills/ folder — skills/makerradar/SKILL.md (when and how an
agent should use MakerRadar, hand-written) plus one folder per skill the site itself serves at
/.well-known/agent-skills/ (find-products, list-a-product, connect-account, check-a-site),
mirrored byte for byte by npm run sync:skills and held in step by CI (--check, daily). A
plugin-aware client, or the Cursor directory, installs it straight from this repo; so does Claude
Code (.claude-plugin/plugin.json + .mcp.json): /plugin install makerradar@<this repo>.
Found it in a directory? Listings on the MCP Registry, Glama, the Claude and ChatGPT directories and friends all point here. A listing that says something this README doesn't is stale: open an issue.
Signing in
makerradar login uses the device flow: it prints a link and a short code, you open the link, sign
in to MakerRadar, check the code and approve. The CLI then gets an agent token for your account
(mr_…). It's never your password or session, and you can revoke it any time in Dashboard › AI
agents.
The token is saved to ~/.config/makerradar/config.json (%APPDATA%\makerradar\config.json on
Windows), readable only by you, and written atomically, so a crash never leaves half a file. Tokens
are kept per API URL, so a staging login never stands in for production. In CI or for an agent,
pass a token with MAKERRADAR_TOKEN instead.
Where a token goes:
MAKERRADAR_TOKENis sent only to the API the environment names (MAKERRADAR_API_URL, or production). An--api-urlpointing anywhere else doesn't get it; pass--tokenfor that API.- A token is never sent over plain
http, except to this machine (localhost,127.0.0.1,[::1]).loginneeds https for the same reason. - An empty or blank variable counts as unset.
For scripts and AI agents
--jsonprints exactly what the API answered ({ "data": … }, plusnextCursoron lists).- On failure, the error goes to stderr. With
--jsonit's JSON:{"error":"not_found","status":404,"message":"…","hint":"…","docs":"…"}. - Exit codes:
0done ·1the API refused, or couldn't be reached ·2a usage mistake (unknown command or option, missing argument). A usage mistake never calls the API. - Text from listings is written by makers. Human output strips terminal control characters and bidirectional overrides, so a listing can't recolour your terminal, move the cursor or fake a link. Treat it as unverified.
Configuration
| Variable | Default | |
|---|---|---|
MAKERRADAR_API_URL |
https://api.makerradar.io/v1 |
The API to use (--api-url) |
MAKERRADAR_TOKEN |
An agent token (--token) for the API above only; otherwise the one login saved |
|
MAKERRADAR_CONFIG |
~/.config/makerradar/config.json |
Where login saves the token |
NO_COLOR |
Plain output (no-color.org) |
Programmatic use
The CLI's API client is exported too:
import { createClient } from "makerradar";
const mr = createClient({ userAgent: "my-script/1.0" });
const { data } = await mr.call("listProducts", { query: { q: "invoicing", limit: 5 } });
Development
npm test # unit tests: a fake API, never the network
npm run test:contract # every endpoint the CLI calls still exists in the live API description
npm run sync:server-json # after a version bump: server.json names the package at its new version
server.json is the MCP Registry entry (mcp-publisher publish sends it; npm version keeps its
package version in step), glama.json names the maintainers for Glama's index, and mcpName in
package.json is how the registry checks the npm package is ours. Pushing a vX.Y.Z tag publishes
both — npm, then the registry — through the release workflow.
No dependencies. The CLI calls only MakerRadar's public, versioned API (/v1). Every endpoint it
uses is listed in src/operations.js, and the contract test checks each one
against the published OpenAPI document every day.
Bugs and ideas: open an issue.
License
MakerRadar MCP questions
How do I add MakerRadar MCP to Claude Code?
Run claude mcp add --env 'MAKERRADAR_TOKEN=<YOUR_MAKERRADAR_TOKEN>' makerradar-cli -- npx -y makerradar in your terminal. Add --scope user to make it available in every project.
How do I add MakerRadar MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
Do I need an account to search for products?
No, you can search the directory, read products, see launches, and draft listings without an account.
How does the CLI handle my login credentials?
The CLI uses a device flow where you approve a code in your browser; it never sees your password and uses an agent token instead.
Can an AI agent manage my product listings?
Yes, by using the account-specific MCP server with an agent token or OAuth, an assistant can update, submit, and manage your listings.