MCP server

Run Phishing Simulations from Your AI Assistant

Connect AutoPhish to any MCP‑compatible assistant. Ask for a campaign draft, a click-rate report or a sender-domain check in plain language, with permissions you control and every call audit-logged.

One server URL. Connect with OAuth or an API key.

Illustrative conversation
YouDraft a phishing simulation for the Sales target list and show me the email before anything goes out.
AI assistantDone. I created a paused draft campaign for the Sales list and rendered the email preview. Nothing is sent until the campaign is activated.

AutoPhish tools used

  • listTargetLists
  • createCampaign
  • previewCampaignEmail

Your awareness program already runs on conversations

Security teams and MSPs lose hours to routine work in dashboards: cloning last quarter's campaign, adding new hires to a list, pulling click and report rates for a meeting. AutoPhish's MCP server lets an AI assistant do that routine work through the same public API your integrations use. You describe the outcome and review the result.

What you can ask your assistant to do

Each example maps to real AutoPhish API tools. Your assistant only sees the tools that the permissions of its connection allow.

Draft campaigns

Create a campaign from a scenario, target list and language, then preview the email before anything is sent.

Example prompt

“Create a draft campaign in German for the Sales list and show me the email preview.”

Tools
ListScenarios
createCampaign
previewCampaignEmail
Permissions
campaigns:read
campaigns:write

Schedule and manage campaigns

Clone earlier campaigns, arm recurring runs and pause a campaign when you need to stop sending.

Example prompt

“Clone our last campaign as a draft, then pause the one that is running right now.”

Tools
cloneCampaignDraft
scheduleCampaign
cancelCampaign
Permissions
campaigns:write

Maintain target lists

Create lists, add or update recipients and import new hires in bulk.

Example prompt

“Add these five new hires to the Onboarding list.”

Tools
listTargetLists
bulkCreateTargets
patchTargetList
Permissions
targets:read
targets:write

Get reports on demand

Pull totals, trends and per-recipient results, and review the phishing emails your employees reported.

Example prompt

“Compare click and report rates for the last 30 days with the 30 days before.”

Tools
getReportSummary
getReportTrends
listPhishingReportedEmails
Permissions
reports:read

Check sender domains

Review SPF, DKIM and DMARC findings for your monitored domains, trigger a fresh check and see which sending domains campaigns can use.

Example prompt

“Which of our monitored domains have open sender-authentication findings?”

Tools
listMonitoredDomains
listDomainFindings
triggerDomainScan
Permissions
integrations:read
integrations:write

Run security scans

Estimate and start security scans, then list open issues and check coverage.

Example prompt

“Estimate a Baseline scan for our main website, then list the open security issues.”

Tools
estimateSecurityScan
startSecurityScan
listSecurityIssues
Permissions
integrations:read
integrations:write

Manage your team

See members and pending invites, invite colleagues and resend invitations.

Example prompt

“Who has a pending invite? Resend the oldest one.”

Tools
listTeamMembers
listTeamInvites
resendTeamInvite
Permissions
team:read
team:write

Review the audit log

See who changed what in your company and export audit entries as CSV.

Example prompt

“Show me the latest audit log entries and export them as CSV.”

Tools
listAuditLogEntries
exportAuditLogEntriesCsv
Permissions
company:read

Some actions stay in the AutoPhish app. PDF report export, logo upload and deleting companies, for example, are not available through MCP.

Connect in a few minutes

AutoPhish runs one remote MCP server. It is built on the public v1 API, so scopes, rate limits and audit logging match the REST API.

  1. Add the server URL

    In your assistant's connector or MCP settings, add the AutoPhish server URL shown below.

  2. Sign in and choose permissions

    Assistants that support OAuth open an AutoPhish sign-in and a consent screen where you pick the companies and the permissions. Clients that accept a custom header can use an API key instead.

  3. Start asking

    Describe what you want in plain language. The assistant calls AutoPhish tools and shows you the result.

Server URL

https://autophish.io/api/mcp

API key configuration

{
  "mcpServers": {
    "autophish": {
      "type": "http",
      "url": "https://autophish.io/api/mcp",
      "headers": {
        "Authorization": "Bearer apk_<prefix>_<secret>",
        "X-Company-Id": "<company-id>"
      }
    }
  }
}

Create the key under Settings → API Keys. For clients that let you set an Authorization header. Field names vary by client, so check your client's documentation for remote MCP servers. Never commit a key to a repository.

Security and control built in

AutoPhish is a security product, so an AI connection is held to the same standard as any other access.

Permissions per connection

On the consent screen you choose the companies you administer and the permissions, just as when creating an API key. Tools outside those permissions are not offered.

Read-only first

Read permissions are pre-selected. Write and sensitive permissions are opt-in, and the consent screen warns you explicitly when you select destructive ones. Nothing changes unless you tick it.

Every call audit-logged

Each tool call is recorded in your audit log with the tool name and the outcome. Arguments and results are never stored there.

Same limits, same isolation

Calls run through the public API, so the same rate limits, validation and company isolation apply. A connection only reaches the companies you selected, and deleting companies is not possible through MCP at all.

Revoke instantly

Manage and remove connected apps under Settings → MCP Server. Their access and refresh tokens stop working immediately.

Admins and owners only

Only company admins and owners can authorize an assistant. If that person loses the role in one company, the connection loses access to that company only.

Built for MSPs and resellers

Managing several customers? One connection can cover several customer companies: you tick them on the consent screen, and the assistant names the company for each action. With the portfolio permissions it can also list your companies and create new ones. A connection only ever reaches the companies you selected.

MCP server FAQ

What is MCP?
The Model Context Protocol is an open standard that lets AI assistants connect to external tools and data. AutoPhish provides a remote MCP server, so a compatible assistant can use AutoPhish capabilities as tools.
Which assistants can I use?
Any assistant or client that can connect to a remote MCP server over HTTP. Assistants such as Claude and ChatGPT support connecting remote MCP servers, as do many developer tools. Availability depends on your assistant, its plan and its settings, so check its documentation. We do not claim certification by or a partnership with any specific product.
Is the MCP server included in my plan?
The MCP server is not sold as a separate add-on. It exposes the same capabilities as the API, and the usual checks of each feature, such as subscription status and token or plan limits, still apply to what the assistant does.
Is my data safe?
Calls run through the public API with the same company isolation and rate limits. You choose the permissions, you can revoke access at any time and every tool call is audit-logged without storing arguments or results. Data that the assistant reads is processed by your AI provider under its terms, so grant only the read permissions you are comfortable sharing.
Can the assistant send emails to my employees?
Yes, but only if you grant campaign write permissions. Campaigns created through MCP are real simulations. The API creates campaigns as paused drafts unless the request activates them, and nothing is sent until a campaign is active. With read-only permissions the assistant cannot create or change campaigns.
Do I need to be a developer?
No. With OAuth you add the server URL, sign in and tick permissions. API-key setups need an MCP client that lets you add an Authorization header.

Run your first phishing test in 10 minutes.

Sign up free — no credit card. Try Pro free for 7 days when you're ready.