Agent

The agent helps you write and maintain documentation

The agent is available on Pro and Custom plans for anyone with access to your dashboard.

The agent creates pull requests with proposed changes to your documentation based on your prompts. When you send a request to the agent, it references your documentation, connected repositories, and Slack messages to create content that follows technical writing best practices and adheres to the Mintlify schema. Access the agent in your Slack workspace or embed it in custom applications with the API.

Use the agent to:

• Write new content based on your prompts, links to pull requests, or Slack threads
• Revise outdated code examples and API references
• Search and update existing content
• Answer questions about your docs and technical writing topics

To get started, add the agent to your Slack workspace and mention it with @mintlify in a channel.

Add the agent to your Slack workspace

If your Slack Workspace Owner requires admin approval to install apps, ask them to approve the Mintlify app before you connect it.

1. Navigate to the agent page of your dashboard.
2. Select the Connect button.
3. Follow the Slack prompts to add the mintlify app to your workspace.
4. Follow the Slack prompts to link your Mintlify account to your Slack workspace.
5. Test that the agent is working and responds when you:
   • Send a direct message to it.
   • Mention it with @mintlify in a channel.

Connect repositories as context

The agent can only access repositories that you connect through the Mintlify GitHub App. Modify which repositories the agent can access in the GitHub App settings.

Embed the agent via API

Use the agent endpoints to create jobs, get a specific job, and get all jobs.

Write effective prompts

Think of the agent as a helpful assistant that needs your guidance to complete tasks. Give it clear instructions and context. More focused tasks are easier to complete, so break down complex projects into smaller steps.

Make your prompts specific and outcome-focused. Generic prompts like @mintlify Improve the onboarding page apply general best practices, but may not improve content in the specific way that you were picturing.

Problem-focused prompts

Try prompts based on outcomes you want your users to achieve or problems that they encounter:

• @mintlify A lot of users have trouble installing the CLI. Review the onboarding page and update the docs so that users can easily install the CLI
• @mintlify Developers keep getting 401 errors when following our authentication guide. Review the auth docs and add clearer examples showing how to properly format the API key
• @mintlify Users are confused about rate limits. Add a clear explanation with examples to the API reference

Content creation prompts

Create new documentation sections or pages with specific requirements:

• @mintlify Create a new troubleshooting page for common database connection errors
• @mintlify Add a section to the quickstart about environment variables with examples for development and production
• @mintlify Write a guide on migrating from v1 to v2 of our API based on this PR: [PR link]

Code example prompts

Generate or update code examples across your documentation:

• @mintlify Add Python and Node.js code examples to the authentication page
• @mintlify Update all API examples to use the new base URL https://api.example.com
• @mintlify Generate a complete code example showing how to handle pagination in our API

Maintenance prompts

Use broad prompts for general content maintenance like fixing typos, updating redirects, or renaming a feature throughout your docs:

• @mintlify Find and fix all typos in the docs
• @mintlify Change all unordered lists to use * instead of -
• @mintlify Replace all instances of "API key" with "access token" throughout the docs
• @mintlify Update all broken links in the deployment section

Review and improvement prompts

Ask the agent to review and improve existing content:

• @mintlify Review the API rate limiting section. We changed limits last month from 100 to 1000 requests per minute
• @mintlify Make the installation guide more concise and easier to follow
• @mintlify Check if all code examples in the SDK documentation are up to date with version 2.0

Specify a subdomain

If you have multiple documentation sites with their own subdomains, include the subdomain parameter in your message to specify which subdomain the agent should work on.

Use the format @mintlify subdomain=<your-subdomain> <your-prompt> to prompt the agent to work on a specific subdomain.

Examples:

• @mintlify subdomain=public-docs Add a new section to the quickstart about inviting collaborators based on this PR: Prompts the agent to update the quickstart only on the public-docs subdomain.
• @mintlify subdomain=customer-docs Update the auth docs for the new authentication method: Prompts the agent to update the auth docs only on the customer-docs subdomain.

Agent workflows

The agent assists with many different documentation tasks. These workflows show some of the ways you can integrate the agent into your documentation process. Try an approach that fits how your team currently works and adapt it to your specific needs.

Iterate on a prompt in a Slack thread

Prompt the agent, then continue to mention it with @mintlify in the same thread to refine and iterate on the pull request that it creates.

Example workflow:

1. Initial prompt: @mintlify Our quickstart page needs a new section on inviting collaborators
2. Refine: @mintlify The new section should be called "Inviting collaborators" and placed after the installation section
3. Add details: @mintlify Add a code example showing how to send an invitation via the API
4. Final touch: @mintlify Add a note about permission requirements for inviting users

Start with the agent, finish manually

Prompt the agent to begin a project, then check out the branch it creates and finish the task in your local environment or the web editor. The agent can help you get started, then you can take over to complete the task.

Example workflow:

1. Prompt: @mintlify Update the quickstart page to include information about inviting collaborators
2. Review the PR the agent creates
3. Check out the branch locally: git checkout mintlify/update-quickstart-collaborators
4. Make additional changes, add screenshots, or refine the content
5. Push your changes and merge the PR

Update docs when merging feature changes

When you merge a feature pull request, share the PR link with the agent to update relevant docs.

Example workflow:

1. Merge a feature PR that adds OAuth support
2. In Slack: @mintlify This PR adds OAuth authentication. Update the authentication docs to include the new OAuth flow with code examples: https://github.com/yourorg/yourrepo/pull/123
3. Review the documentation PR the agent creates
4. Merge the documentation updates

Real-world example:

@mintlify This PR adds webhook signature verification: 
https://github.com/acme/api/pull/456

Update the webhooks documentation to:
- Explain how signature verification works
- Add code examples in Python and Node.js
- Include troubleshooting tips for common verification errors

Generate release notes from a pull request

Prompt the agent with a specific pull request to generate release notes or changelog updates based on the commit history.

Example workflow:

1. Complete a sprint with multiple merged PRs
2. Prompt: @mintlify Generate release notes for version 2.1.0 based on these PRs: [PR link 1], [PR link 2], [PR link 3]
3. Review and refine: @mintlify Add a migration guide section for the breaking changes
4. Merge the changelog update

Generate code examples

Prompt the agent to generate code examples for features throughout your docs or on specific pages.

Example workflow:

1. Prompt: @mintlify Add code examples for the new batch API endpoint in Python, Node.js, and cURL
2. Review the examples
3. Refine: @mintlify Add error handling to the Python example
4. Merge the updates

Real-world example:

@mintlify Our authentication page only has cURL examples. 
Add examples in Python, Node.js, Ruby, and Go showing:
- How to authenticate with an API key
- How to refresh tokens
- How to handle authentication errors

Review existing content

Prompt the agent to review existing content for technical accuracy, style, grammar, or other issues.

Example workflow:

1. After a major API update: @mintlify Review all API documentation and update any references to the old v1 endpoints to use v2 endpoints
2. Review the changes: @mintlify Also check for any outdated parameter names
3. Merge the updates

Real-world example:

@mintlify We changed our rate limits from 100 to 1000 requests per minute last month. 
Review all documentation and update:
- Rate limit numbers
- Code examples showing rate limit headers
- Troubleshooting guides mentioning rate limits

Respond to user feedback

Prompt the agent with feedback from your users to make focused updates to your docs.

Example workflow:

1. Receive user feedback: “Step 3 in the setup guide is confusing”
2. Prompt: @mintlify Users are getting confused by step 3 in the setup guide about environment variables. Make it clearer by adding examples and explaining what each variable does
3. Review and merge the improvements

Real-world example:

@mintlify Multiple users reported that our WebSocket documentation doesn't explain 
how to handle reconnections. Add a section covering:
- Automatic reconnection strategies
- Exponential backoff examples
- How to detect connection state

Sync documentation with Slack threads

Reference Slack threads to provide the agent with full context from team discussions.

Example workflow:

1. Have a detailed discussion in Slack about a new feature
2. Prompt: @mintlify Based on this thread, create documentation for the new caching feature: [Slack thread link]
3. The agent reads the entire thread and creates comprehensive documentation
4. Review and merge

Automate with the API

Integrate the agent into your existing automation tools to automatically update documentation when code changes occur, trigger content reviews, or sync documentation updates across multiple repositories.

Example workflow:

1. Set up a GitHub Action that triggers when PRs are merged to main
2. Use the agent API to automatically create documentation updates
3. Review the automated PRs and merge

Learn how in the Auto-update documentation when code is merged tutorial.