MCP Server Terms and AI Assistant Integration

Version: 2026-05-29.v1

Last updated: June 17, 2026

1. What is the MCP Server?

Our MCP (Model Context Protocol) server allows you to connect AI assistants (Claude Desktop, ChatGPT, and other MCP clients) to your Semly Pro account. This enables AI assistants to:

• Access your SEO data (keywords, content gaps, audit results)
• Generate content based on your projects
• Publish content to your connected CMS platforms
• Run analyses and retrieve insights

The MCP server acts as a bridge between the AI assistant and your Semly Pro account, executing actions on your behalf based on the AI's requests.

2. Scope of Access

2.1 Organization-Scoped Access

MCP connections are scoped to your organization. The AI assistant can only access data and perform actions for projects within the organization that granted access. It cannot access other users' organizations or data.

2.2 Permission Levels

When connecting an AI assistant, you grant one of these scopes:

• Read-only: AI can retrieve insights, data, and reports but cannot generate content or publish
• Content generation: AI can generate content and run analyses but cannot publish to CMS
• Full access: AI can generate content, run analyses, AND publish to your connected CMS platforms

Important: Choose the minimum scope necessary. Full access allows the AI to publish content to your live websites.

2.3 Revoking Access

You can revoke MCP access at any time through:

• Semly Pro dashboard: Settings → Integrations → MCP Connections → Revoke
• Your AI client's MCP settings (disconnect Semly Pro server)

Revocation takes effect immediately. The AI assistant will no longer be able to access your Semly Pro account.

3. Your Responsibilities

3.1 Review AI Outputs

CRITICAL: You are solely responsible for reviewing all AI-generated content before it is published to your CMS. This includes:

• Verifying factual accuracy (AI models can hallucinate)
• Checking for copyright infringement
• Ensuring compliance with your industry regulations
• Reviewing tone, style, and appropriateness
• Adding required AI content labels per EU AI Act Article 50 (effective Aug 2, 2026)

Publishing unreviewed AI content that violates laws, infringes copyrights, or contains false information is a violation of our Acceptable Use Policy.

3.2 Secure Your AI Client

You must:

• Keep your AI assistant client (Claude Desktop, ChatGPT app) secure
• Not share your MCP connection with unauthorized users
• Revoke access if you lose control of your AI client device
• Monitor AI activity through Semly Pro's MCP usage logs

3.3 Comply with AI Provider Terms

When using the MCP server, you must also comply with:

• Anthropic's Terms of Service (for Claude)
• OpenAI's Terms of Use (for ChatGPT)
• Terms of any other MCP client you use

Violations of AI provider terms may result in loss of MCP access.

4. Data Processing and Privacy

4.1 Data Shared with AI Providers

When the AI assistant calls MCP tools, data is transmitted to the AI provider (Anthropic, OpenAI, etc.):

• Your prompts and instructions to the AI
• Data returned by MCP tools (SEO insights, generated content, etc.)
• Context from your conversation with the AI

This data is processed under the AI provider's privacy policy:

• Anthropic Privacy Policy
• OpenAI Privacy Policy

No Training on Your Data: Anthropic and OpenAI have committed not to train their models on data submitted through their business APIs. However, standard API terms apply for safety and abuse monitoring.

4.2 Sensitive Data Warning

Do not include sensitive personal data in AI prompts. This includes:

• GDPR Article 9 special categories (health, biometric, genetic data)
• Financial credentials or payment information
• Passwords or API keys
• Personal information about individuals without their consent

If you must process sensitive data, ensure you have:

• Explicit consent from data subjects
• Appropriate legal basis under GDPR/CCPA
• Executed data processing agreements with AI providers

4.3 MCP Usage Logs

We log MCP tool calls for:

• Security and abuse detection
• Usage monitoring and plan limit enforcement
• Debugging and support

Logs include: timestamp, tool name, organization ID, success/failure status. We do not log the full content of prompts or tool responses.

5. Plan Limits and Rate Limiting

5.1 Plan Quotas

MCP usage counts toward your plan limits:

• Content generation: Same limits as dashboard content generation
• SEO audits: Same limits as dashboard audits
• API calls: Subject to plan-specific rate limits

5.2 Rate Limiting

To prevent abuse, we enforce rate limits:

• Tool calls: 100 calls per minute per organization
• Content generation: Plan monthly limits apply
• Publishing: 10 publishes per hour (prevents accidental mass publishing)

If you exceed rate limits, the MCP server will return an error to the AI assistant. Persistent abuse may result in temporary or permanent suspension of MCP access.

6. Service Availability and Support

6.1 Beta Status

The MCP server is currently in beta. This means:

• Features may change or be discontinued
• Uptime and reliability are not guaranteed
• Support is provided on a best-effort basis
• Breaking changes may occur with limited notice

Once MCP moves to general availability (GA), we will provide standard SLAs and support.

6.2 No Warranty

The MCP server is provided "as is" without warranty of any kind. We do not warrant that:

• The MCP server will be available, reliable, or error-free
• AI-generated outputs via MCP will be accurate or fit for purpose
• MCP will be compatible with all AI clients or MCP implementations

6.3 Support

MCP support is provided through:

• Documentation: docs.semlypro.com/mcp
• Email support: mcp-support@semlypro.com
• Community: MCP-specific support channel (coming soon)

7. Prohibited Uses

7.1 General Prohibitions

All prohibitions in our Acceptable Use Policy apply to MCP usage, including:

• Illegal activity
• Harmful or misleading content
• Spam and abuse
• Copyright infringement
• Privacy violations

7.2 MCP-Specific Prohibitions

Additionally, you may not:

• Reverse-engineer: Attempt to extract or replicate MCP tool implementations
• Build competing services: Use MCP to build a competing SEO platform
• Credential sharing: Share MCP access tokens or connection details
• Automated abuse: Use MCP for mass automated content generation beyond reasonable use

8. Liability and Indemnification

8.1 Our Liability Limitations

We are not liable for:

• AI-generated content created via MCP that infringes copyrights or contains false information
• Damage caused by publishing unreviewed AI content to your CMS
• Loss of data or interruption of service
• Actions taken by AI assistants on your behalf

Our liability is limited as specified in our Terms of Service Section 6.

8.2 Your Indemnification

You agree to indemnify us for claims arising from:

• Content you publish via MCP to your CMS
• Your violation of AI provider terms while using MCP
• Your processing of personal data via MCP in violation of GDPR/CCPA
• Your violation of these MCP Terms or our Acceptable Use Policy

9. Changes and Termination

9.1 Changes to MCP Terms

We may update these MCP Terms from time to time. For material changes, we will:

• Increment the version number
• Notify you by email
• Provide 30 days notice before changes take effect

Continued use of MCP after changes take effect constitutes acceptance.

9.2 Termination of MCP Access

We may terminate your MCP access if:

• You violate these MCP Terms or our Acceptable Use Policy
• You exceed rate limits or abuse the service
• Your Semly Pro subscription is canceled or terminated
• We discontinue the MCP server (with 90 days notice)

10. Contact

Related legal documents:
Terms of Service | Privacy Policy | Acceptable Use Policy | Data Processing Agreement