# AI Agents on WordPress.com: Automating Content Creation and Management
WordPress.com has expanded capabilities for AI agents, allowing them to generate posts, edit metadata, manage comments, and structure content via tags and categories. All operations are initiated by natural language commands through the interface. This paves the way for fully automated site creation without manual content intervention.
The platform handles 20 billion page views monthly and 409 million unique visitors. Support for the MCP (Model Context Protocol), introduced in the fall, enables LLM context transfer for accessing content, settings, and analytics from external tools like Claude Desktop, Cursor, or VS Code.
AI Agent Features
AI agents now perform not just data reading but also active actions:
- Creating posts, landing pages, and "About Us" pages;
- Handling comments: approval, replies, editing;
- Managing taxonomy: creating, renaming, restructuring categories and tags;
- SEO optimization: adjusting alt text, captions, titles;
- Generating drafts with meta descriptions, tags, and categories.
Changes are logged in the activity journal. AI takes into account the site's design — colors, fonts, spacing, block patterns — before generating content. All publications require owner approval; AI-generated content is saved as a draft by default.
To activate, go to wordpress.com/mcp and enable permissions. Supported MCP clients: Claude, Cursor, ChatGPT, and others.
MCP and Context for LLMs
The MCP protocol standardizes context exchange between applications and large language models. On WordPress.com, it allows AI to extract site data (posts, settings, statistics) and apply changes. This simplifies integration with IDEs and chatbots, where developers describe tasks in natural language.
Example workflow: the agent analyzes the site theme, generates a post in the style of templates, adds SEO tags, and sends for approval. The activity log provides auditing of all operations.
Additional AI Tools
Since February, a built-in AI assistant for images based on Google Gemini has been available. Features:
- Generating new images with parameters (aspect ratio, style);
- Editing existing visuals;
- Integrating the Generate Image button into the interface.
This complements textual content, automating the full material production cycle.
Key Points
- AI agents require explicit approval for publications, minimizing risks;
- MCP provides secure site access without direct API;
- Logging in the activity journal simplifies control and reverting changes;
- Integration with site design ensures consistency;
- Support for external MCP clients expands the tool ecosystem.
Prospects for Developers
For mid/senior specialists, this means automating routine tasks: from generating landing pages to SEO optimization. Integrating MCP into workflows (VS Code + Cursor) speeds up prototyping. Test on staging sites, monitor the log for debugging.
— Editorial Team
No comments yet.