Run your app presence
from any MCP client

AppLander ships a built-in MCP server. Connect Claude, Cursor, or VS Code with one command and a browser login, then manage everything by typing a sentence.

$ claude mcp add --transport http applander https://applander.io/mcp

That is the whole configuration. No API token, no header, no secret on your clipboard.

Create a free account Already on AppLander? Connect it

Included on every plan, free included — 126 tools covering every resource in your account.

Four tasks, three dashboard sections,
one sentence

The dashboard is not going anywhere. But some days you do not want to click through it.

Before

Ten minutes of clicking

1 Open the FAQ editor, write three entries, save each one.
2 Switch apps, open the status page, post an incident.
3 Switch sections again, paste in last night's changelog draft, publish.
4 Open tickets, filter, close the five you already resolved.

With MCP

One prompt

> Add an FAQ about password resets to FitTrack, post a payment-degraded incident on BudgetBuddy's status page, publish the v2.3 changelog entry, and close all resolved tickets.

Your assistant picks the right tools, fills in the parameters, and reports back. Same permissions, same plan limits, same validations as the dashboard.

Connect it in under two minutes

Every client authenticates the same way — through your browser. There is nothing to generate, copy, or paste.

Add the server

Claude Code

$ claude mcp add --transport http applander https://applander.io/mcp

Cursor, Claude Desktop, Claude.ai

Add a custom connector with type HTTP and URL https://applander.io/mcp

VS Code

Create .vscode/mcp.json — safe to commit, because it holds no secrets:

{
  "mcpServers": {
    "applander": {
      "type": "http",
      "url": "https://applander.io/mcp"
    }
  }
}

Approve it in your browser

In Claude Code run /mcp, pick applander, then Authenticate. Your browser opens on AppLander:

Sign in, if you are not already.
Review the consent screen — it names the client asking and the account it will reach.
Click Authorize and go back to your editor. It now says connected.

This is the standard MCP OAuth 2.1 handshake — discovery, dynamic client registration, PKCE. Your client stores and refreshes the token itself, so you do this once.

Check that it works

> List my apps on AppLander.

You should get your apps back with their names, statuses, and prefix IDs. That is it — you are connected.

Things to ask it

Real prompts, and the tools your assistant reaches for to answer them.

Start here

Launch a whole app presence from one sentence

> Create a new app on AppLander called FitTrack. Write the landing page from my store listing, generate a privacy policy and terms covering GDPR and CCPA, add the ten FAQs a fitness app always gets asked, set up the support page, and publish it.

One prompt, and every page your store listing asks for exists at a real URL — landing page, privacy policy, terms, support, account deletion. What used to be an afternoon of forms is a paragraph you type once.

create_app create_landing_section create_privacy_policy create_terms_of_service create_faq update_app

Batch FAQ writing

"Add five FAQs to FitTrack: restoring purchases, cancelling a subscription, syncing with Apple Health, enabling notifications, and contacting support. Make the answers specific to a fitness app."

create_faq list_faq_categories

Incident response

"Post a major incident on BudgetBuddy's status page titled 'Database maintenance — read-only mode', then add an update saying full service is back within 30 minutes."

create_incident create_incident_update

Changelog from your commits

"Create a changelog entry for FitTrack 2.4.0 released today: dark mode, fixed calorie maths for custom meals, better workout history charts. Publish it."

create_changelog_entry publish_changelog_entry

Support triage

"Show me all open tickets for FitTrack. Close the ones about the sync bug — we fixed it in 2.4.0 — and reply to the rest saying we are looking into it."

search_tickets bulk_update_tickets

Localisation

"Duplicate all of BudgetBuddy's English content into a new Spanish language."

create_language duplicate_language_content

Legal upkeep

"Update FitTrack's privacy policy to mention Firebase Analytics for crash reporting, and publish it with today's effective date."

update_privacy_policy publish_privacy_policy

Store readiness

"Is FitTrack ready to submit to the App Store and Google Play? Tell me what is missing."

get_submit_readiness

Landing page edits

"Add a testimonials section to FitTrack's landing page with the three reviews I just pasted, and move it above the pricing section."

create_landing_section reorder_landing_sections

What your assistant can reach

126 tools, one per operation, so reads and writes stay separable and your client can ask before it changes anything.

Area	Reads	Writes	Plan
Apps	List, inspect, check submit readiness	Create, update, delete, set icon and OG image	Free
Landing pages	List sections, read the section schema	Create, update, reorder, delete sections; add and remove images	Free
Features	List and inspect	Create, update, reorder, delete	Free
Privacy, terms, cookies	Read current text, list versions, diff two versions	Create, update, publish, roll back a version	Free
FAQs	List categories and entries	Create, update, delete categories and entries	Free
Support tickets	List, search, read the message thread	Update, bulk update, reply, delete	Free
Deletion requests	List and inspect GDPR requests	Add operator notes, advance the workflow	Free
Languages	List locales	Add a locale, duplicate all content into it	Free
Changelog	List and inspect entries	Create, update, publish, unpublish, delete	Pro
Guides	List categories and guides	Create, update, delete	Pro
Articles	List and inspect the per-app blog	Create, update, publish, unpublish, delete	Pro
Status page	List components and incidents	Create and update components, open, update, and resolve incidents	Pro
Canned responses	List and inspect	Create, update, delete	Pro
Analytics	Page view summaries, top pages, top and unanswered FAQ searches	—	Pro

Connecting is free on every plan. Pro-only tools prompt you to upgrade when used — exactly like a gated button in the dashboard. See plans

You stay in control

Giving an AI assistant write access to your public pages is a real decision. Here is exactly what it can and cannot do.

No secrets to manage

You never generate, copy, or paste a token. Connecting is OAuth 2.1 with PKCE — the same handshake Claude Code, Cursor, and ChatGPT already speak.

You approve every client

Each new client shows a consent screen naming it and your account before it gets any access. Nothing connects silently.

Short-lived, rotating tokens

Access tokens expire after an hour and refresh automatically. Refresh tokens rotate on use, so a stolen one dies the moment your real client refreshes.

A real off switch

Settings → AI clients lists every connected client with its last-active time. Disconnect revokes all of its access immediately.

Your account only

Every query is scoped to your account. It is architecturally impossible to reach another user's data through your connection.

Plan limits still apply

Limits are re-checked on every request. Your assistant cannot publish a second app on Free or bypass a Pro gate — it gets a clear message and an upgrade link.

Requests are capped at 120 per minute, and every tool call is logged with the client name, the tool, its duration, and whether it succeeded.

Works with your client

Streamable HTTP with OAuth 2.1 — no per-client token setup anywhere.

Claude Code

Claude Desktop

Claude.ai

Cursor

VS Code

Windsurf

Warp

Zed

ChatGPT

Anything else that speaks the Model Context Protocol over Streamable HTTP will work too. Use a current build — older releases had incomplete OAuth support.

Questions? Answers.

Do I need to generate an API token? +

No. You connect with one command and approve the client in your browser — there is no token to create, copy, or paste. AppLander still offers API tokens for its REST API, but the MCP server does not use them.

Do I need to be a developer? +

No. If you can describe what you want in a sentence, you can use it. Your assistant works out which tool to call and what to pass it.

Can Claude accidentally delete my app? +

Destructive tools like delete_app are separate from every read tool, so your client asks before calling one. You see a confirmation prompt before anything is deleted.

Does this cost extra? +

No. Connecting is available on every plan, free included. Changelog, guides, articles, status pages, canned responses, and analytics are Pro features, so using those tools prompts you to upgrade — exactly like the dashboard.

How do I disconnect a client later? +

Go to Settings → AI clients, find the client, and click Disconnect. Its access is revoked immediately. You can also remove it on the client side — for Claude Code, claude mcp remove applander.

What happens when I hit a plan limit? +

Your assistant gets a plain message explaining which plan is needed, plus a link to upgrade. No silent failures, no confusing error codes.

Is my data sent to Anthropic? +

Your data flows through your assistant's context during the conversation, subject to that vendor's data policies. AppLander's MCP server sends data only to the client you connected — it stores no conversations and shares nothing with third parties.

I set this up with an API token and it broke. What changed? +

The endpoint moved to browser-based OAuth. Remove the old entry and re-add it with no header: claude mcp remove applander, then claude mcp add --transport http applander https://applander.io/mcp. Your REST API tokens are unaffected.

Want the long version, with worked examples? Read the deep dive on the blog.

Stop clicking. Start typing.

Create a free account, connect your client in two minutes, and run your whole app presence from wherever you already work.

Get started free See all features

Run your app presencefrom any MCP client

Four tasks, three dashboard sections, one sentence