What is the agent?

Learn how the Mintlify agent researches, plans, and writes documentation, then opens pull requests with proposed changes for your team to review.

The agent is an AI tool that creates pull requests with proposed changes to your documentation based on your prompts. When you send a request, the agent:

• Researches: Searches and reads your existing documentation, any connected repositories, relevant context, and the web.
• Plans: Creates a structured task list for complex documentation work.
• Writes: Generates or updates content following writing standards and best practices.
• Validates: Runs Mintlify CLI checks to ensure documentation builds correctly.
• Publishes changes: Opens a pull request with proposed changes for review, or commits directly to your deployment branch, depending on your review process setting.

Use the agent to:

• Write new content based on your prompts, pull requests, Slack threads, or file attachments.
• Update existing documentation for new features or API changes.
• Process and include images, diagrams, and other files from Slack attachments.
• View images in your docs, connected repositories, and attachments to describe them, write alt text, or check that screenshots match your content.
• Search and revise code examples and API references across your docs.
• Reference source code from any repository that has the Mintlify GitHub App installed.
• Automate documentation maintenance with automations.
• Answer questions about your docs and technical writing topics.
• Address code review feedback to maintain documentation quality.
• Capture knowledge from Slack conversations and pull requests before it gets lost.
• Navigate and coordinate changes across multiple files and directories efficiently.

By default, the agent opens pull requests attributed to the Mintlify bot. To attribute pull requests to you, connect your GitHub account on the My profile page of the dashboard.

By default, the agent opens a pull request for every change so your team can review it before it ships. If Mintlify hosts your project, the agent commits changes directly to your deployment branch so changes publish immediately.

For projects connected to your own Git provider, click Configure Slack on the Automations page of your dashboard, then set Review process in Slack settings:

• Create a pull request: The agent commits to a new branch and opens a pull request against your deploy branch. Use this option when you want a human to review changes before they publish.
• Push to main: The agent commits directly to your deploy branch, and changes publish as soon as the agent finishes. Use this option when you trust the agent to publish without review, for example on a staging deployment.

The agent can only access repositories that you connect through the Mintlify GitHub App. Configure which repositories the agent can access in the GitHub App settings. For GitLab, use Git Settings.

If your Git provider restricts access with an IP allowlist, Mintlify can route Git requests from agent sandboxes through a static IP address.

Contact Mintlify about the project that you want to enable static agent egress for. Once enabled, the static address applies to Git operations from the agent and automations, including cloning, fetching, and pushing connected repositories.

The agent works asynchronously through pull requests, but you can also use AI coding tools like Cursor or Claude Code locally for fast, iterative edits. Install the Mintlify skill and connect the Model Context Protocol (MCP) server so your editor has the same context the agent uses.

To let an external agent edit your content directly through MCP tool calls, connect it to the Mintlify MCP server.

Install the Mintlify skill for context on project structure, components, and documentation best practices:

npx skills add https://mintlify.com/docs

Add the Mintlify Docs MCP server for documentation search:

https://mintlify.com/docs/mcp

Add the Mintlify MCP server for dashboard access and content management:

https://mcp.mintlify.com

• skill.md
• Ask agent
• Customize agent behavior