Kaminari Ad + AI Agents

Manage ad verification with one prompt

Connect Kaminari Ad to Cursor, Claude Desktop, or any MCP-compatible agent. Trigger scans, write policies, monitor alerts — in plain English. 108 tools, no dashboard needed.

Install in Cursor Download for Claude Desktop

Why an MCP server

Stop clicking through dashboards

Tell your AI assistant what you need. It calls the right Kaminari Ad tool — and gives you a structured answer back.

Scan creatives in seconds

Describe the URL and target GEO. The AI picks the device profile, queues the scan, and returns the verdict — no dashboard clicks.

Investigate verdicts in plain English

"Why did scan #abc fail?" The AI fetches the scan, evidence, redirect chain, and tag matches in one back-and-forth.

Author policies as you speak

"Flag any creative claiming 100% returns." The AI writes a custom rule, dry-runs it on past scans, then deploys.

Monitor alerts without the dashboard

"What alerts came in overnight?" Get a grouped list. Escalate or resolve in one command.

Cost attribution on demand

"How much did campaign X spend this week?" Get usage broken down by campaign, org, or day — without building a report.

Your API key stays on your machine

Local stdio install passes the key directly to Kaminari Ad — no MCP-side storage. Hosted mode uses Bearer auth, no caches.

What can you do?

Five workflows that take one prompt

Each example wires together several MCP tools — your AI assistant orchestrates the calls.

Scan a creative across multiple countries

"Scan https://example.com/banner across the top 20 GEOs on real iPhones. Return any that flagged cloaking."

list_emulators → create_bulk_scans → list_scan_tags

Build and test a custom rule

"Write a rule that fires when a landing page mentions 'guaranteed returns'. Test it against my last 200 scans before I activate it."

create_custom_rule → test_custom_rule

Schedule a campaign with Slack alerts

"Set up a recurring campaign that re-scans my top 50 advertisers every 6 hours. Send any violation to #ad-quality Slack."

create_campaign → set_campaign_alert_overrides

Cost report by campaign

"Show me check count and spend by campaign for the last 7 days."

get_usage_summary → list_usage

Replay a failed webhook delivery

"Delivery #d-abc failed yesterday. Replay it now and tail the next 3 attempts."

replay_webhook_delivery → list_webhook_deliveries

Installation guide

Install in under a minute

Click 'Install in Cursor'

The button above (or at the top of this page) opens Cursor with a pre-filled mcpServers entry. If Cursor isn't installed, the page stays put with the deeplink URL you can paste manually.

Paste your API key

In the config form Cursor shows, paste your KAMINARI_AD_API_KEY — get one at app.kaminari.ad → Settings → API Keys.

Save and start chatting

Open any chat in Cursor and try "List my recent scans" or "Show open alerts".

Alternative: edit config file manually

~/.cursor/mcp.json

{
  "mcpServers": {
    "kaminari-ad": {
      "command": "npx",
      "args": ["-y", "@kaminari-ad/mcp"],
      "env": {
        "KAMINARI_AD_API_KEY": "your-key"
      }
    }
  }
}

Advanced · No local install

Hosted HTTP transport

For serverless agents, CI pipelines, or clients without a local Node runtime — point at the hosted endpoint with a Bearer header. Same 108 tools, no Node install.

{
  "mcpServers": {
    "kaminari-ad": {
      "url": "https://mcp.kaminari.ad/mcp",
      "headers": {
        "Authorization": "Bearer your-key"
      }
    }
  }
}

Tool catalogue

108 tools across 15 domains

One tool per /api/v1 call, with MCP behaviour annotations (readOnly / destructive / idempotent) so your client can warn before risky actions.

Account

14 tools

Org, users, roles, labels, API keys — invite, revoke, transfer ownership.

Scans

13 tools

Queue scans (one or bulk by GEO), inspect verdicts, re-check, cancel, pull screenshots, creative HTML, video and VAST XML.

Campaigns

10 tools

Create, schedule, run, archive, list runs — recurring verification jobs.

Campaign Groups

10 tools

Folder-level operations: run all, cancel all, pause/resume schedules.

Runs

3 tools

Inspect run-level counters and the scans they contain; cancel.

Custom Rules

6 tools

Author plain-English detection rules. Test-rule (dry-run) before deploy.

Custom Taxonomies

7 tools

Define your own content categories, parse them from plain text, restore deleted ones.

Policy Sets

11 tools

Group rules + tags into policies, request approval, set the default, attach to campaigns.

Alerts

4 tools

List violation alerts, escalate/resolve/dismiss one or in bulk, stats by status.

Webhooks

11 tools

Billing

4 tools

Balance, usage, plan limits, ledger transactions.

Invoicing

2 tools

Invoice list with status and amounts, invoice PDF download.

Alert Notifications

5 tools

Slack / Telegram / Email / Webhook destinations and per-campaign routing.

Reference data