Skaler

npx add-mcp 'https://skaler.app/api/mcp'

Two minutes

Connect your AI assistant

Let your assistant set it up for you, or add it yourself.

Option 1: ask your assistant

Paste this into Claude, ChatGPT or Cursor. Coding agents add the server themselves; chat apps tell you exactly where to click.

Paste into Claude, ChatGPT or Cursor

Set up the Skaler MCP server for me.
Server URL: https://skaler.app/api/mcp
Transport: streamable HTTP. Auth: OAuth (sign in with my Skaler account).
Setup guide: https://skaler.app/mcp
If you can add MCP servers yourself (e.g. Claude Code: \`claude mcp add --transport http skaler https://skaler.app/api/mcp\`), do it. Otherwise give me the exact steps for the app I'm using.

Option 2: add it yourself

Pick your app and follow the steps.

claude.ai and the Claude desktop app

1. 1Open Claude → Settings → Connectors.
2. 2Click “Add custom connector”, name it Skaler and paste the server URL.
3. 3Click Connect, sign in to Skaler and allow access.
4. 4In a chat, open the tools menu and make sure Skaler is turned on.

https://skaler.app/api/mcp

On Team and Enterprise plans an owner adds the connector once for the organization.

Connection settings

The details your client asks for

Server URL

https://skaler.app/api/mcp

Sign-in

OAuth 2.1 with PKCE and dynamic client registration. You sign in with your Skaler account.

Scope

email

Who can use it

Any Skaler account. The demo ad account is free; your own ad accounts need a Skaler plan.

Tools

18: 14 read, 4 write

Charts

Data tools return an interactive chart in clients that support MCP Apps.

What it can do

Everything Skaler knows about your ads

Top creatives

Ranked ads with metrics, Skaler creative scores, AI tags and thumbnails.

AI tag performance

Which hooks, angles, personas and formats win in your account.

Account summary

KPIs vs the previous period, daily trend, ready-made Slack update.

Audience & placement

Age, gender, country, platform and format breakdowns.

Transcripts & analysis

The transcripts and AI analyses you already made in Skaler.

Saved reports

Run any of your Skaler reports with live data.

Launch from the chat

Create campaigns and launch ads

Describe the test you want. Your assistant builds it in Skaler's launcher, shows you the summary, and launches only when you say go.

Launch options

get_launch_options

Your campaigns (CBO or ABO), ad sets, Pages, Instagram accounts and call-to-action types.

Create campaign

create_campaign

A new Meta campaign: Sales or any other objective, ad set budgets (ABO) or a campaign budget (CBO). Paused by default.

Create launch draft

create_launch_draft

Copies one of your ad sets with new videos or images from links, plus copy, link and schedule. Nothing reaches Meta yet; review it in Skaler.

Launch draft

launch_draft

Uploads the creatives and creates the ad sets and ads. Paused unless you choose active or a scheduled start.

Launch status

get_launch_status

What was created, with Meta ad and ad set IDs, and why any ad failed.

Paused until you say otherwise

• Campaigns, ad sets and ads start paused.
• Your assistant asks before each step.
• A launch repeats the ad count and ad set name you approved.

Prompt library

Things to ask first

Copy one into your assistant once Skaler is connected.

Winners

This week's winning ads

Using Skaler, show my top 10 ads by ROAS from the last 7 days with at least $100 spend. For each, give the hook, format and one line on why it's working.

Winners

Ads to kill

Using Skaler, find ads from the last 14 days that spent the most with the worst CPA. Rank them and tell me which to pause first and why.

AI tags

Which hooks win for me

Using Skaler, rank my hook types by ROAS for the last 30 days. Show spend share for each and tell me which hook to make more of.

AI tags

Personas and angles

Using Skaler, compare my personas and messaging angles over the last 30 days. Which persona + angle combinations should I test next?

Reporting

Weekly Slack update

Using Skaler, write my weekly ad account update for Slack: spend, revenue, ROAS and CPA vs last week, the top 3 ads, and one action for next week.

Reporting

Run my saved report

List my saved Skaler reports, run the one I use most for the last 7 days and summarise what changed.

Audience

Who actually buys

Using Skaler, break down ROAS by age, gender and placement for the last 30 days. Where should I shift budget?

Creative briefs

Brief from my winners

Using Skaler, take my 5 best ads of the last 30 days, look at their transcripts and AI tags, and write 3 new ad briefs that reuse what works for my brand.

Launch

New creative test campaign

Using Skaler, create a paused ABO test campaign called "Creative test – November" with the Sales objective. Show me what you'll create before you do it.

Launch

Launch new videos, paused

Using Skaler, copy my best ad set into that campaign as 3 new ad sets at $10/day, one per video: <video link 1>, <video link 2>, <video link 3>. Keep the copy and link from the ad set. Make a draft, show me the summary, and launch it paused once I say go.

Launch

Did my launch work?

Using Skaler, show the status of my last launch: which ads were created, their Meta ad IDs, and why any failed.

Tools reference

All 18 tools

What each tool does and whether it reads or writes. Your assistant picks the tool; you never have to name one.

Your ad account

Read-only analytics for the Meta ad accounts you connected in Skaler, plus a demo account.

Tool	What it does	Effect
get_workspace Get Skaler workspace	Your plan, ad accounts (always including the demo account), brands and AI-tag categories. Assistants call it first.	Read
query_creatives Query creative performance	Ranked ads with spend, ROAS, CPA, CTR, hook and hold rate, Skaler's four creative scores, AI tags and thumbnails. Filter and group by any of them.	Read
get_ai_tag_performance AI tag performance	Which hooks, angles, personas, formats and offers win in the account, with spend share for each.	Read
get_account_summary Account performance summary	KPIs against the previous period, a daily trend and the top ads. Can return a ready-made Slack or markdown update.	Read
get_breakdowns Audience & placement breakdowns	Spend or ROAS split by platform, placement, format, country, age and gender.	Read
get_ad_insights Saved transcript & AI analysis	The transcript and AI analysis already created in Skaler for one ad.	Read
list_reports List saved reports	Your saved Skaler reports (sample reports on the demo account).	Read
get_report Run a saved report	Runs one of your saved reports with live data.	Read

Competitor research

Read-only lookups in Skaler's library of public Meta ads. No ad account needed.

Tool	What it does	Effect
search_winning_ads Search winning Meta ads	Finds winning ads in any niche from Skaler's ad library, with run time and scaling signals.	Read
spy_on_brand Spy on a competitor's ads	What one brand is running and scaling right now: active ads, formats, landing pages and longest runners.	Read
analyze_ad_creative Analyze a competitor ad	Breaks down one ad: hook, angle, structure, copy and why it likely works.	Read
list_tracked_brands List tracked competitor brands	The brands you follow in Skaler's Brand Spy.	Read

Launching

These create things in your Meta ad account. Your assistant asks before each call, and everything starts paused unless you say otherwise.

Tool	What it does	Effect
get_launch_options Get launch options	Your campaigns (CBO or ABO), ad sets, Pages, Instagram accounts and call-to-action types.	Read
create_campaign Create Meta campaign	A new, empty Meta campaign with any objective, ABO or CBO.	Write to Meta Paused by default
create_launch_draft Create launch draft	Builds a draft in Skaler's launcher from creative links: which ad set to copy, copy, link, budget and schedule. Returns a summary and a link to review it.	Write Skaler draft only, nothing reaches Meta
launch_draft Launch draft	Uploads the creatives and creates the ad sets and ads from an approved draft. Needs the ad count and ad set name from the draft summary.	Write to Meta Spends money only if you ask for active or a scheduled start
get_launch_status Get launch status	What a launch created, with Meta ad and ad set IDs, and why any ad failed.	Read

Feedback

Tool	What it does	Effect
send_feedback Send feedback to Skaler	Sends a bug report or idea to the Skaler team when you ask your assistant to.	Write A message to Skaler support

Limits, safety and privacy

What it can and can't do

Limits and usage

Tool calls

500 per user in any 24 hours

Date range

Up to 31 days per question. 7 to 14 days answer fastest.

create_campaign

20 a day

create_launch_draft

50 a day

launch_draft

20 a day

send_feedback

10 a day

Monthly plans

MCP calls are included.

Credit-based plans

1 credit per data or competitor lookup. Workspace, saved transcripts, report lists and launch tools are free.

No plan

The demo account works in full. Competitor lookups return up to 3 ads.

Safety boundaries

• Analytics and competitor tools only read. They can't change anything in Meta.
• Campaigns, ad sets and ads are created paused unless you ask for active or give a start time.
• A draft launches only after you've seen its summary. The launch call has to repeat the ad count and ad set name from that summary, and separately confirm if anything would go live.
• It never pauses, edits or deletes existing campaigns, ad sets or ads.
• It never sees your Meta login or access tokens.
• It only reaches ad accounts you connected in Skaler.

Full details in the privacy policy.

You approve every launch

Analytics are read-only. Launches start paused and only run after you confirm.

Your login, your data

Secure OAuth sign-in. It only sees accounts you connected.

Disconnect any time

Account Settings → Connected AI apps.

Troubleshooting

When something looks off

The assistant says the data is still loading

Skaler loads a new date range from Meta day by day. Ask the same question again in about a minute, or use a shorter range. No credits are used while it loads.

It only shows the demo account

Your own ad accounts appear once you have a Skaler plan and a Meta ad account connected in Skaler. Sign in to the connector with the same Skaler account.

It keeps asking me to sign in

Remove the connector in your AI app, add it again with the server URL above and approve access on the Skaler screen.

Skaler doesn't appear in ChatGPT

Custom MCP servers need Developer mode: Settings → Apps & Connectors → Advanced. Then pick Skaler from the + menu in a new chat.

No chart, only numbers

Charts need a client that supports MCP Apps. Elsewhere the assistant gets the same numbers and can build a chart from them. Every result also links to the same view in Skaler.

“Date range too long”

Each question covers up to 31 days. For longer history, open the linked report in Skaler.

“Daily limit reached”

The limit is 500 calls in a rolling 24 hours, so it frees up as older calls age out.

“Not enough Skaler credits”

On credit-based plans each data lookup uses 1 credit. Check your balance in Skaler under Settings → Plan. The demo account stays free.

A launch created fewer ads than expected

Ask for the launch status. It lists every ad with its Meta ID and the reason any ad failed, such as a creative link that couldn't be downloaded.

Still stuck? Email support@skaler.app, or ask your assistant to send feedback to Skaler and it reaches the team directly.

MCP directories

Where Skaler is listed

Frequently asked questions

Everything you need to know about Skaler MCP

Ask away, I don't bite.

What is the Skaler MCP server?

MCP (Model Context Protocol) lets AI assistants use outside tools. Skaler's MCP server gives Claude, ChatGPT, Cursor and other MCP clients read access to your Skaler data: Meta ad performance, creative scores, AI tags, breakdowns, transcripts and saved reports. It can also create campaigns and launch ads through Skaler's launcher.

Which AI assistants work with it?

Claude (claude.ai and desktop), ChatGPT with developer mode, Claude Code, Codex, Cursor and any client that supports remote MCP servers with OAuth.

Can it change my ads or spend money?

Only the launch tools write to Meta, and only when you approve each step. Your assistant asks before every call that creates something. Campaigns, ad sets and ads start paused unless you ask for active or set a start time, and a draft only launches after you've seen its summary. It never pauses, edits or deletes existing ads. Everything else is read-only, and it never sees your Meta login or tokens.

How do launches work?

Your assistant builds a draft in Skaler's launcher: which ad set to copy, the new creatives (from links), copy, link, budget and schedule. You get a summary and a link to review or edit the draft in Skaler. When you say go, Skaler uploads the creatives to Meta and creates the ads, retrying temporary Meta errors. Ask for the launch status any time to see the new ad IDs.

How much does it cost?

It's included in monthly Skaler plans; on credit-based plans each data lookup uses 1 credit. Anyone can try it on the demo account for free.

Why is the first answer for a long date range slow?

Meta only returns so much data per request, so Skaler loads date ranges day by day and caches them. If that takes a while, your assistant is told to try again a minute later. Each question covers up to 31 days; 7 to 14 days answer fastest. For longer history, open the report in Skaler.

How do I disconnect it?

In Skaler go to Account Settings → Connected AI apps and click Disconnect. You can also remove the connector in your AI app.

Setup instructions for AI assistants

Skaler MCP server URL: https://skaler.app/api/mcp. Transport: streamable HTTP. Auth: OAuth 2.1 with dynamic client registration; the user signs in with their Skaler account and approves access.

All 18 tools: get_workspace, query_creatives, get_ai_tag_performance, get_account_summary, get_breakdowns, get_ad_insights, list_reports, get_report, search_winning_ads, spy_on_brand, analyze_ad_creative, list_tracked_brands, get_launch_options, create_campaign, create_launch_draft, launch_draft, get_launch_status, send_feedback.

Launch tools: get_launch_options, create_campaign, create_launch_draft, launch_draft, get_launch_status. They write to the user's Meta ad account: confirm each call with the user, keep everything paused unless the user asks otherwise, and call launch_draft only after the user approved the draft summary.

• Claude: Open Claude → Settings → Connectors. Click “Add custom connector”, name it Skaler and paste the server URL. Click Connect, sign in to Skaler and allow access. In a chat, open the tools menu and make sure Skaler is turned on.
• ChatGPT: Open ChatGPT → Settings → Apps & Connectors → Advanced, and turn on Developer mode. Back in Apps & Connectors, click Create, name it Skaler and paste the server URL. Choose OAuth, click Create, then sign in to Skaler and allow access. Start a new chat and pick Skaler from the + menu.
• Claude Code: Run the command below in your terminal. Start Claude Code and run /mcp, pick skaler and sign in to Skaler in the browser. Command: claude mcp add --transport http skaler https://skaler.app/api/mcp
• Codex: Run the command below in your terminal. Run codex mcp login skaler and sign in to Skaler in the browser. Command: codex mcp add skaler --url https://skaler.app/api/mcp
• Cursor: Open Cursor Settings → MCP → Add new MCP server (or edit ~/.cursor/mcp.json). Add the server below, save, and click Connect to sign in to Skaler. Config: { "mcpServers": { "skaler": { "url": "https://skaler.app/api/mcp" } } }

Set up the Skaler MCP server for me. Server URL: https://skaler.app/api/mcp Transport: streamable HTTP. Auth: OAuth (sign in with my Skaler account). Setup guide: https://skaler.app/mcp If you can add MCP servers yourself (e.g. Claude Code: `claude mcp add --transport http skaler https://skaler.app/api/mcp\`), do it. Otherwise give me the exact steps for the app I'm using.