mintlify · mintlify · May 9, 2026
diff --git a/assistant/configure.mdx b/assistant/configure.mdx
@@ -85,6 +85,15 @@
 
 By default, the assistant allows overages. You can disable overages to avoid incurring additional costs for usage beyond your package. If you reach your credit allowance with overages disabled, the assistant is unavailable until your credit allowance resets. If you allow overages, each credit beyond your allowance incurs an overage charge, but occasional overages may be cheaper than upgrading to a higher package depending on your usage.
 
+### View usage breakdown
+
+The [Usage](https://dashboard.mintlify.com/settings/organization/usage) page breaks down credit consumption across your AI features so you can see where credits are being spent and investigate spikes.
+
+- **Daily consumption** chart: A stacked bar chart of daily credit usage broken down by product (Assistant, Agent, Workflows, MCP, and Others). Totals for each product appear in the legend below the chart, with their share of overall credit usage.
+- **Date range**: Switch between **Last 7 days**, **Last 30 days**, **Last 90 days**, and **Last 12 months** to adjust the time window for the chart.
+- **Product filter**: Choose **All products** to compare every product, or pick a single product (Assistant, Agent, Workflows, MCP, or Others) to focus the chart on that product only.
+- **Activity log**: A chronological list of credit-consuming events across all AI features. Each row shows the timestamp, the product and source (for example, Web chat, Slack agent, or Workflow), and the credits used. Select **Load more** to page through older events.
+
 ### Change your assistant package
 
 Packages determine your monthly credit allowance and pricing.

diff --git a/changelog.mdx b/changelog.mdx
@@ -9,14 +9,18 @@

  ## Expanded MCP content access

  Authenticated [MCP servers](/ai/model-context-protocol) now return more of your content to authorized users. Authenticated users can search public pages plus any pages they have permission to access based on their user groups, instead of being limited to public pages only. [Client credentials](/ai/model-context-protocol#client-credentials) return all public pages and authenticated pages that aren't restricted to specific groups.

  The Mintlify MCP `list_nodes` tool now supports filtering by `parentId` (with optional `recursive` traversal), multiple node types, and additional division scopes including `dropdown`, `anchor`, `product`, and `item`. Results are paginated through an opaque cursor. See the [Mintlify MCP](/ai/mintlify-mcp) documentation for more details.

  ## Assistant credit packages
 
   The [assistant](/assistant) billing experience has been updated to use credit-based packages across all plans, with consistent terminology in the dashboard and usage controls. No changes to existing credit balances or usage.
 
+  ## Credit usage breakdown
+
+  The [Usage](https://dashboard.mintlify.com/settings/organization/usage) page now breaks down credit consumption by product so you can see where credits are spent across Assistant, Agent, Workflows, MCP, and other AI features. A new daily consumption chart visualizes usage over the last 7 days, 30 days, 90 days, or 12 months, with optional filtering by product. An activity log below the chart lists individual credit-consuming events, including the source (for example, Web chat, Slack agent, or Workflow) and credits used. See [View usage breakdown](/assistant/configure#view-usage-breakdown) for more details.
+
   ## AsyncAPI nested groups
 
   The `asyncapi` property now supports [nested groups](/api-playground/asyncapi-setup#examples-with-nested-groups), so you can generate channel pages inside a subsection of a broader API group or combine multiple AsyncAPI specifications under a shared parent.
@@ -25,7 +29,7 @@

  - [Slack agent](/agent/slack) now requests permission requests to run third-party action such as sending an email or creating a ticket.
  - The [editor](/editor) now lets you drag images and videos from the navigation tree directly onto a page to embed existing media inline.
  - Mermaid diagrams in the editor now support a fullscreen mode for easier editing of complex diagrams.
  - The custom domain dashboard now displays both verification `TXT` records (`_acme-challenge` and `_cf-custom-hostname`) with live verification status, so you can confirm DNS is correct before pointing your `CNAME` at Mintlify. See [custom domain](/customize/custom-domain) for more details.
  - TLS certificate provisioning for [custom domains](/customize/custom-domain#automatic-tls-provisioning) is now handled directly by Mintlify.
  - GitLab-connected deployments can now revalidate their Git source from the [Git settings](https://dashboard.mintlify.com/settings/deployment/git-settings) page by clicking the **Active** badge, useful when branch options or configuration appear stale.
@@ -39,7 +43,7 @@

  Workflows have been rebuilt with a new [dashboard experience](https://dashboard.mintlify.com/products/workflows?tab=workflows).

  - **Optimized workflow templates**: Mintlify maintains prompts for common content creation and maintenance tasks. Turn on these workflows and let the agent handle the rest. Mintlify routinely improves the prompts so that you don't have to.
  - **Quicker setup**: Toggle on a workflow and adjust any optional settings. No more YAML configuration or iterating on Markdown prompts.
  - **Custom workflows**: In addition to built-in workflows, you can create your own workflows.

@@ -47,7 +51,7 @@

  ## Editor

  The [editor](https://dashboard.mintlify.com/editor) is redesigned for better collaboration and improved stability. The editor now supports real-time collaboration, automatic saving across devices, and improved syncing.

  See the [editor](/editor) documentation for more details.

@@ -67,7 +71,7 @@

  Ask the agent in [Slack](/agent/slack) to set up any third-party integration that it should use as context. Prompt the agent to connect to a service like GitHub, Jira, Notion, Google Drive, Confluence, or any other service that you use.

  The agent uses [Composio](https://docs.composio.dev/toolkits) to support a broad catalog of third-party services, including issue trackers, knowledge bases, CRMs, and developer tools. See the [Composio toolkits catalog](https://docs.composio.dev/toolkits) for the full list of supported services.

  ## Section-level search boost

@@ -75,9 +79,9 @@

  ## Improvements

  - Control whether OpenAPI-generated reference pages are included in your [Markdown export](/ai/llmstxt) and `llms-full.txt`
  - Search improvements: matched query terms highlighted in search results, search indexes content inside alt text, tag-based search filters are case-insensitive, and snippets no duplicate page titles.
  - Banner components support more types and per-user dismissibility.
  - Authenticated navigation now stays sticky while scrolling.
  - The [Slack agent](/agent/slack) can now clone any repository the Mintlify GitHub App has access to and is aware of every repo in your org when answering questions.
  - The Assistant CSV export and API now include resolution status for each conversation.
@@ -86,7 +90,7 @@

  ## Bug fixes

  - Fixed copying of personalized code blocks so the rendered values are copied instead of the templated source.
  - Fixed prefetching incorrectly attempting to fetch external links.
  - Fixed prose styling for the [Tabs](/components/tabs) component.
  - Fixed offset alignment for blocks placed inside a [Column](/components/columns).
@@ -99,7 +103,7 @@

 </Update>

 <Update label="April 24, 2026" tags={["New releases", "Improvements"]} rss={{ title: "Preview Deployment API, Git onboarding redesign, web editor improvements, and more" }}>

  ## Preview Deployment API

@@ -107,15 +111,15 @@

  ## Git onboarding and settings redesign

  The Git settings page has been redesigned with a new layout. Mintlify-hosted customers migrating to their own GitHub or GitLab repository now get a step-by-step flow that walks through the full migration process.

  ## Web editor improvements

  The web editor now shows the full Git diff for the current branch, including changes already pushed. Review everything modified in one place before publishing or opening a pull request.

  Changes now persist locally, so edits made while offline are preserved and synced automatically when your connection returns.

  - **Emoji picker**: Type <kbd>:</kbd> to open an emoji picker and insert emojis inline.
  - **Accordion reordering**: Drag and drop to rearrange accordions inside an [accordion group](/components/accordions) without editing MDX.
  - **Editable color component**: Update [color](/components/color) swatches directly in the editor without editing MDX.
  - **SVG file upload**: Upload SVG files from the editor configuration view.
@@ -127,7 +131,7 @@
  ## Improvements

  - Pages expose a `data-current-path` HTML attribute for easier custom scripting and analytics targeting.
  - Hidden pages now respect the `seo.indexing: all` setting so you can still index specific pages that are excluded from navigation.

  ## Bug fixes

@@ -170,12 +174,12 @@
  - Pasting valid MDX components into the editor now preserves component structure instead of converting to plain text.
  - Long sidebar navigation titles now hyphenate automatically to prevent overflow.
  - The editor now resolves [`$ref` references](/organize/settings#split-configuration-with-%24ref) in split configuration files.
  - The [Slack agent](/agent/slack) system prompt has been improved for more accurate and helpful responses.
  - `llms.txt` now includes blockquote summaries for better page descriptions.
  - Custom `skill.md` files in your repository are now detected correctly during deployment.

  ## Deprecations
  - File-based workflows are deprecated. Create workflows from the dashboard instead. Existing workflows from files continue to work, but new workflows must be created in the dashboard.

  ## Bug fixes

@@ -211,11 +215,11 @@

  ## Multi-skill support

  Serve multiple [skill files](/ai/skillmd) from a `.mintlify/skills/` directory. Each subdirectory with a `SKILL.md` file is discovered and served automatically, so you can organize skills by feature or domain.

  ## Slack agent improvements

  - **Multiple deployments**: Organizations with multiple documentation sites can now use the [Slack agent](/agent/slack) across all of them. The agent prompts you to choose a deployment on first use, and your preference is saved per Slack account.
  - **Read-only mode**: The Slack agent now classifies messages and responds in read-only mode for questions that only need information retrieval, reserving write access for requests that require documentation changes.

  ## Client credentials for authenticated MCP
@@ -227,19 +231,19 @@
  - The [`mint new`](/cli/commands#mint-new) command now supports creating projects from pre-defined templates with the `--template` flag.
  - The CLI now auto-refreshes expired access tokens, so you no longer need to re-run `mint login` after a session expires.
  - The [assistant](/assistant/index) is now available in `mint dev` local previews.
  - CLI UX improvements: grayed-out login URLs for clarity, standardized log formatting, and `mint status` now shows the CLI version.
  - OpenAPI `x-group` extension support for organizing API endpoints into custom groups.
  - Horizontal rules (`<hr />`) are now supported in MDX content.
  - [Workflows](/workflows) now surface in the dashboard inbox for easier tracking. The workflow list also displays PR names for quicker identification.
  - The `mint export` command now includes API reference pages generated from OpenAPI specifications.
  - The slash command menu in the editor has been redesigned with improved organization and discoverability.
  - Improved the UI of the editor publish flow and page creation interface.
  - Improved abuse prevention for the assistant.
  - Added Finnish as a supported localization language.

  ## Bug fixes

  - Fixed custom heading IDs not being processed before MDX parsing, which could cause anchors to break on certain pages.
  - Fixed the SEO and metadata audit workflow template from incorrectly modifying API endpoint pages.
  - Fixed deprecated label text wrapping in the dashboard.
  - Fixed inline formatting (bold, italic, code) not applying correctly in the editor in certain contexts.
@@ -251,7 +255,7 @@

 </Update>

 <Update label="April 3, 2026" tags={["New releases", "Improvements"]} rss={{ title: "get_page MCP tool, password-protected previews, SAML group role mappings, and more" }}>

  ## get_page MCP tool

@@ -267,10 +271,10 @@

  ## Improvements

  - Redesigned the editor configuration settings page with an improved layout and autosave support.
  - Pages auto-generated from OpenAPI specifications now appear in the web editor as read-only entries.
  - Video support in [live preview](/editor/live-preview). Embedded mp4, webm, and mov files now render correctly during local development.
  - [Workflows](/workflows) can now be disabled and re-enabled from the dashboard.
  - Assistant analytics now show usage broken down by source (web vs API) in the dashboard.
  - Authenticated MCP search now correctly filters results based on the user's authentication groups."
  - Slack agent responses now show tool call summaries instead of raw tool calls for clearer information of what the agent is doing.
@@ -308,13 +312,13 @@

  Use inline AI editing to generate, rewrite, and restructure content directly in the visual editor. Select text and choose from preset actions or type a custom instruction. Changes stream in as a highlighted diff so you can review before applying.

  ## Composable `docs.json`

  Split large configurations into smaller files using [`$ref` references](/organize/settings#split-configuration-with-%24ref). Each `$ref` points to a separate JSON file resolved at build time.

  ## `data-active` CSS selectors

  Sidebar items, table of contents entries, and navigation dropdowns now expose a stable [`data-active` attribute](/customize/custom-scripts#css-selectors) for custom CSS targeting. Use selectors like `.toc-item[data-active]` or `#sidebar-content li[data-active]` to style active states without relying on internal class names.

  ## Agent-skills discovery endpoint

@@ -347,7 +351,7 @@
  - Fixed padding for tables inside tabs.
  - Fixed Frame component hint icon vertical alignment.
  - Fixed navigation item badge wrapping.
  - Fixed API key being erased during prefill examples.
  - Fixed linked code snippets in cards not being clickable.

 </Update>
@@ -360,7 +364,7 @@

  ## New contextual menu options

  The [contextual menu](/ai/contextual-menu) now includes options for **Google AI Studio**, **Devin**, and **Windsurf**. Users can open any documentation page directly in these tools with the page content preloaded as context.

  ## Self-hosted GitLab configuration

@@ -376,7 +380,7 @@
  - Workflow templates updated with improved prompts.
  - Workflow run history now viewable from the dashboard with the ability to monitor status and close PRs.
  - Slack notifications for workflows now configurable from the dashboard.
  - Editor performance improvements: fixed typing latency and persistence blocking in large repositories, virtualized file lists for repositories with many files.
  - Card layout shift on initial page load eliminated.
  - Dashboard analytics date-picker responsiveness improved.
  - Search now uses a unified endpoint with improved performance and consistency.
@@ -459,9 +463,9 @@

  The assistant now also builds context from the page a user is on and indexes entire OpenAPI specifications.

  ## Default Vale configuration updates

  The default [Vale configuration](/deploy/ci#configuration) now parses MDX as MD to avoid fragility with JSX. Previously, Vale returned false positive errors on any JSX inside of MDX components. Now, Vale parses MDX as MD to avoid these errors with tokenIgnores and blockIgnores for MDX and JSX elements that Vale should skip.

  ## Improvements

@@ -476,7 +480,7 @@
  ## Bug fixes

  - Fixed OAuth and JWT redirects for authenticated docs.
  - Fixed cursor jumping when pasting text in the editor.
  - Fixed keyboard shortcuts for canceling assistant messages.
  - Fixed branch names with underscores not working in the editor.
  - Fixed link border colors in callout components.
@@ -532,7 +536,7 @@

  ## Improvements

  - New search functionality in the web editor makes it easier to find and navigate to files across your documentation.
  - Drag to add table rows and columns directly in the editor for faster table creation and modification.
  - The API playground now supports audio response examples, making it easier to document audio-related API endpoints.

@@ -563,7 +567,7 @@

  The [contextual menu](/ai/contextual-menu) now includes an option to add a copyable command to install your docs MCP server.

  ## `root` in groups

  Navigation groups now support a [`root`](/organize/navigation#root-page) property to set a landing page for the group. Clicking the group header navigates to the root page instead of expanding the group.

@@ -651,7 +655,7 @@
  - Fixed a rendering loop that could cause performance issues with the Tabs component.
  - Fixed the password reset button not working.
  - The AI assistant no longer appears when printing pages.
  - Fixed the cursor being invisible in dark mode.
  - Fixed a bug that could strip style attributes in MDX.
  - Fixed an issue where page "last modified" dates updated incorrectly during global config changes.
  - Fixed authentication issues affecting index pages from properly rendering.
@@ -661,7 +665,7 @@

 <Update label="January 23, 2026" tags={["New releases", "Improvements"]} rss={{ title: "skills.md, audit logs, and markdown export" }}>

  ## `skill.md` auto-generation

  Documentation sites now auto-generate a `skill.md` file based on your API references, guides, and features to give AI agents a comprehensive skills file with context on how to help people use your product. `skill.md` replaces the `install.md` approach with more structured capabilities and wider adoption. If you have a custom `skills.md` file at the root of your documentation repository, it replaces the auto-generated version. See [skill.md](/ai/skillmd) for more information.

@@ -671,19 +675,19 @@

  ## Markdown export for agents

  Documentation pages now serve raw markdown when the `Accept: text/markdown` header is sent, which enables better AI agent consumption of content.

  ## Improvements

  - Combined authentication, partial authentication, and personalization into a single authentication experience. See [Authentication setup](/deploy/authentication-setup) for more information.
  - Hidden pages now automatically receive `noindex` meta tags for better SEO control.
  - Added markdown support for image captions.

 </Update>

 <Update label="January 16, 2026" tags={["New releases", "Improvements"]} rss={{ title: "install.md, live preview in the web editor, and more" }}>

  ## `install.md` auto-generation
  Documentation sites now auto-generate an `install.md` file that aggregates installation-related pages like quickstarts, getting started, installation, and setup guides for AI agents. If you have a custom `install.md` file in your documentation repository, it takes priority over the auto-generated version.

  ## Web editor improvements
@@ -693,7 +697,7 @@

  ## Branch search for Git integrations

  Added branch search functionality for GitHub and GitLab integrations so that it is easier to find and switch between deployment branches in the Git settings of your dashboard.

  ## `mintlify validate` CLI command

@@ -791,7 +795,7 @@
  - Updated [publishing workflow](/editor/branching-and-publishing) to support one-click publishing or creating pull requests.
  - Working in the web editor no longer requires you to sign in with GitHub.

  ## Onboarding improvements

  - Choose between a Mintlify-hosted documentation repository or using your own repository while signing up.
  - Create a documentation site without connecting a GitHub account.
@@ -810,9 +814,9 @@
 <Update label="December 12, 2025" tags={["New releases", "Improvements"]} rss={{  title: "Agent suggestions and Q&A bots" }}>
  ## Agent suggestions

  The agent can now monitor Git repositories and suggest documentation updates. When user-facing code changes are detected, suggestions appear in the agent panel. Suggestions include the relevant pull request and a list of proposed documentation updates. Add suggestions as context for the agent to create pull requests or dismiss them.

  ## Q&A bot in Slack

  Add a bot powered by the assistant to your [Slack workspaces](/ai/slack-bot). The bot answers questions in real-time based on your documentation, so it provides accurate, up-to-date answers to your community.

@@ -826,7 +830,7 @@

  - **Insights** renamed to **Analytics** in the dashboard.
  - Added ability to delete a deployment from the dashboard.
  - The `Tooltip` component has been redesigned with improved styling and now supports an optional `headline` prop for adding a title to your tooltips.
  - The `mint a11y` command now supports `--skip-contrast` and `--skip-alt-text` flags, allowing you to run specific accessibility checks in your CI/CD pipeline. Use `--skip-contrast` to check only for missing alt text, or `--skip-alt-text` to check only color contrast ratios.

  ## Bug fixes
@@ -835,11 +839,11 @@
  - Fixed heading levels 5 and 6 rendering in the web editor.
  - Fixed schema page examples display in API reference pages.
  - Fixed image upload paths in monorepo setups.
  - Fixed an issue where the incorrect request payload value was being selected first in the API playground.
  - Removed file extensions from default image titles when uploading.
  - Fixed double underline styling on banner links.
  - Fixed loading states and date range values for analytics page.
  - Folders in the web editor now properly maintain their open/closed state when other folders are opened or closed.
  - Fixed long API page titles overflowing instead of wrapping.

 </Update>
@@ -861,7 +865,7 @@

  You can now display "last modified" timestamps on all pages in your documentation. Enable this feature by setting the [`timestamp`](/organize/settings-seo#metadata) flag in your `docs.json` settings.

  When enabled, each page will automatically show when it was last updated, helping readers understand how current the content is.

 </Update>

@@ -877,8 +881,8 @@

  ## Assistant improvements

  - Fixed assistant submission to prevent empty messages from being sent.
  - Fixed accidental message submission during IME composition (for languages like Japanese, Chinese, Korean).
  - Better behavior when adding code snippets to assistant.

  ## Bug fixes and reliability
@@ -886,7 +890,7 @@
  - Fixed handling of complex Mermaid diagrams.
  - Fixed hash handling for URLs and table of contents to properly highlight the active page when accessing the root path with an index page.
  - Added ability to disable 404 page recommendations via configuration.
  - Fixed a bug in the API playground where boolean `false` and numeric `0` values in query parameters would be filtered out and not appear in the generated API request.

 </Update>

@@ -902,7 +906,7 @@
 <Update label="December 1, 2025" tags={["New releases"]} rss={{ title: ".mintignore support", description: "Exclude files and directories from your published documentation" }}>
  ## .mintignore support

  - Added `.mintignore` file support to exclude specific files and directories from being processed and published to your documentation site.
  - Uses the same pattern syntax as `.gitignore` for familiar configuration.
  - Excluded files don't appear in published docs, aren't indexed for search, and aren't accessible to visitors.

@@ -923,7 +927,7 @@
  - Fixed issue where response section would disappear when switching between endpoints with different response codes. The playground now properly resets to the first available response code when navigating to a new endpoint.
  - Fixed double "Authorization" header display in HTTP Basic authentication schemes. The authentication section now shows a cleaner interface with proper field labels and descriptions.
  - Improved handling of `oneOf` schemas in nested objects and arrays. Complex API schemas with multiple type options now render correctly as dropdowns instead of tabs in appropriate contexts.
  - Fixed missing path parameters that were defined at the endpoint level in OpenAPI specs. Parameters specified above individual operations are now properly included in both the main page and playground.
  - Enhanced response example generation to include all required properties, even when examples only specify some fields. This ensures response examples are complete and accurate.
  - Fixed missing prefill examples in object arrays. Array items with example data now properly populate in the playground with correct indexing.
  - Improved description padding and spacing throughout the API playground for better readability and visual consistency.
@@ -934,7 +938,7 @@

  ## CLI improvements

  - Added error message for users running `mint dev` on Node.js versions below 20.17. Users are guided to upgrade to an LTS version.

 </Update>

@@ -953,14 +957,14 @@
  - Security inputs defined at the operation level in OpenAPI specs now properly override top-level security definitions
  - Fixed issue where header inputs marked as security parameters were incorrectly included in generated code examples
  - Improved handling of `deepObject` style parameters in API field displays
  - Better error messages when invalid URLs are provided in API configurations
  - Improved UI for API playground with better spacing, layout, and loading

  ## Navigation and UI enhancements

  - Tab hover indicators now have improved styling for better visual feedback
  - Long words now wrap properly to prevent horizontal scrolling
  - Step component now handles cases where no step title is provided
  - Logo section spacing improved with better padding
  - Fixed heading IDs to properly handle question marks and special characters for anchor links
  - Icon component now accepts `className` prop for custom styling
@@ -975,7 +979,7 @@

  ## Component and styling enhancements

  - Images can now be zoomed with improved overlay styling for both light and dark modes
  - Fixed keyboard shortcut styles inside callouts for dark mode
  - Improved OG image generation with Google Fonts support and better truncation logic
  - Improved UI for tables with better alignment, spacing, and readability
@@ -1002,22 +1006,22 @@
 <Update label="October 31, 2025" tags={["Improvements"]} rss={{ title: "Weekly Updates", description: "API playground enhancements, table improvements, performance optimizations, and bug fixes" }}>
  ## Assistant improvements

  - **Starter questions for assistant are here! You can add your own within the dashboard at [Assistant --> Settings](https://dashboard.mintlify.com/mintlify/mintlify/products/assistant/settings)**
  - Assistant insights quality has been improved with default spam protection for abusive keywords and JSON queries.

  ## API playground improvements

  - The API playground now automatically fills in default values for parameters, which saves time when testing endpoints.
  - Issues with `allOf`, `oneOf`, and nested object schemas that previously caused rendering problems or incorrect field displays have been fixed, and complex API schemas now render correctly with all fields visible and properly structured.
  - Enum and nullable field types now display accurate labels in the playground, which makes it clearer what values are accepted and whether fields are optional.
  - When working with nested objects in API requests, the playground now shows proper descriptive labels for each field instead of generic placeholders, which makes it easier to understand what data to provide.
  - Enum const tables and API field displays now render with improved formatting and alignment, which makes it easier to scan available options and understand field requirements.
  - API playground modals now display rich MDX-formatted descriptions at the top, which gives better context for endpoints with formatted text, links, and code examples.

  ## Analytics changes

  - Deferred analytics loading: PostHog and LogRocket now load after documentation content, which improves initial page load times and Core Web Vitals scores without affecting analytics data.
  - Conditional session recording: PostHog session recording can now be enabled conditionally based on user properties or page context for more control over when recordings happen.

  ## Visual improvements

@@ -1041,7 +1045,7 @@
  ## Insights improvements

  - Improved insights page with fixed date selectors for "today" and "yesterday"
  - Enhanced feedback table with improved search functionality
  - Added loading state for categories table

  ## Assistant improvements
@@ -1126,7 +1130,7 @@
 <Update label="October 11, 2025" tags={["Improvements"]} rss={{ title: "Weekly Updates", description: "API playground improvements, analytics enhancements, web editor updates, and bug fixes" }}>
  ## Assistant and Agent AI features

  - Added list_pull_requests and list_commits tools for agent such that it can document a date range of PRs or multiple PRs at once
  - Upgraded agent and assistant to Claude Sonnet 4.5
  - Improved assistant search to query docs in parallel for faster assistant responses
  - Fixed conversation length counting to exclude tool calls
@@ -1142,7 +1146,7 @@
  - Removed background from code blocks in editor for cleaner appearance
  - Fixed branch creation workflow to correctly navigate to newly created branches
  - Added deployment history pagination
  - Added cursor pagination to preview deployments table
  - Fixed preview deployments to only return preview environment deployments
  - Improved dashboard UI/UX with better spacing, input handling, and layout fixes
  - Added better handling for links in dashboard assistant chat
@@ -1158,14 +1162,14 @@
  ## Bug fixes and reliability

  - Fixed duplicate base paths and links in 404 recommended pages
  - Auto-populated og:url tag with appropriate canonical URL
  - Added canonical tag for better SEO
  - Fixed code block overscroll issues
  - Made search engines ignore preview deployments for SEO purposes
  - Fixed referrer tracking to use domain instead of full URL
  - Fixed images always becoming MDX block elements (kept inline images inline)
  - Removed comments in raw markdown pages such that you can use TODO comments without them leaking to users
  - Fixed directory reading support for read_external_files

  ## Component and styling

@@ -1181,7 +1185,7 @@
  - **Accessibility CLI command**: Run `mint a11y` to test your docs for accessibility issues
  - **Video responses in API playground**: API playground now supports displaying video response types
  - **API playground prefill**: New option to prefill API playground examples for faster testing
  - **Unauthenticated feedback**: For sites using partial authentication, users can now submit feedback without being authenticated
  - **Shiki theme support**: Shiki themes are now supported for code block styling
  - **Twoslash for code blocks**: TypeScript and JavaScript code blocks now support Twoslash for inline type information

@@ -1220,7 +1224,7 @@

  - Added support for Romanian and Czech languages in the documentation interface
  - Enhanced localization capabilities with complete translation coverage for new languages
  - Improved language selector functionality across all themes

  ## UI and user experience improvements

@@ -1256,7 +1260,7 @@
  - Fixed GitLab integration issues with larger repositories
  - Improved dashboard loading performance with optimized deployment queries
  - Fixed SVG path rendering issues in documentation
  - Fixed keyboard navigation in search and chat functionality
 </Update>

 <Update label="September 13, 2025" tags={["New releases", "Improvements"]} rss={{ title: "Weekly Updates", description: "AI 404 suggestions, assistant web search with external sources, security enhancement, and reliability fixes" }}>
@@ -1276,10 +1280,10 @@

  ## Web editor and deployments

  - Security enhancement added to editor such that only users with `write permissions` for the connected git hosting repository can make changes
  - Fixed a bug where preview deployments wouldn't work for branches with `=` in the name
  - Fixed a bug where long branch names would overflow modals on preview deployment creations
  - Quality of life improvement where email query parameter will prefill the input on signup invitations
  - Fixed a bug where copying a page from the context menu was not working on safari

  ## API playground and navigation
@@ -1292,7 +1296,7 @@
  - Fixed a bug where favicons were not showing up in search engines by serving them from the same URL as the documentation site itself for each respective site
  - Fixed a bug where youtube embeds would flash in and out on load
  - Fixed a bug where expanding the feedback menu to include written responses would cause layout shift with the table of contents
  - Fixed a bug where text would leak above the topbar on the maple theme when a dismissed the notification banner
  - Enhanced the Maple and Willow themes by adding login/logout buttons to the sidebar for easier access

  ## Analytics and exports
@@ -1309,10 +1313,10 @@

  ## Navigation and quality of life improvements

  - Quality of life improvement to ignore trailing slashes and double slashes so you don't have to worry about getting them exactly right in your docs.json
  - You can now add a `noAnchor` attribute to your `h1-6` HTML tags to avoid them having an anchor link
  - The Palm theme now has a floating language selector in the bottom left similar to Stripe's approach
  - Added a new field to the docs.json called [`drilldown`](https://mintlify.com/docs/navigation#enable-auto-navigation-for-groups) which allows you to control whether or not users are navigated to the first page in a navigation group when the group is expanded
  - Quality of life improvement to make nested ordered lists alternate styles from decimal/Roman to alpha

  ## Bug fixes and reliability
@@ -1321,34 +1325,34 @@
  - Fixed a bug where Google was indexing raw `*.md` files because they were missing a `x-robots-tag noindex` header
  - Fixed a bug with OAuth on protected docs where it wouldn't redirect you back to your starting page once you completed the flow successfully
  - Fixed a bug on previews of auth protected docs where you weren't able to see the entire navigation bar
  - Bug fixes to how SVGs are handled with our new image CDN

  ## Component and styling enhancements

  - Added a new CSS selector for custom styles on `SidebarNavGroupDivider`
  - New regression tests for MDX defined API pages that have security defined on them will ensure a greater degree of stability

  ## Performance improvements

  - Performance improvement by moving the KaTeX CSS from cdnjs to our own CDN on Cloudfront for reduced latency
 </Update>

 <Update label="August 23, 2025" tags={["Improvements"]} rss={{ title: "Weekly Updates", description: "Image layout fixes, auth enhancements, mobile nav improvements, and web editor stability" }}>
  ## Image handling improvements

  - **Major improvement**: Images no longer cause layout shift by default, even when width and height attributes aren't specified—automatic sizing prevents content jumping during page loads
  - All static files in your repository (PDF, TXT, XML, etc.) are now automatically uploaded and served when you deploy, providing complete asset coverage

  ## Web editor and deployment enhancements

  - Fixed branch creation workflow in web editor to correctly navigate to and stay on newly created branches
  - Enhanced merge conflict dialog with proper escape functionality, no more page reloads required to dismiss conflicts
  - Optimized update workflow performance by cache-invalidating only changed pages during partial updates, reducing deployment times

  ## Authentication and navigation improvements

  - New support for authentication on custom subpaths, if you serve docs at `https://yourdomain.com/docs`, authentication now works seamlessly
  - Fixed sidebar display bug that incorrectly showed when only one link was configured
  - Comprehensive mobile navigation overhaul: centered buttons with proper margin/padding, improved spacing in dropdown menus, removed unnecessary dividers and margins for empty sections, and fixed Maple theme gap/padding issues

  ## Component and styling fixes
@@ -1360,7 +1364,7 @@

  - Enhanced logging system for update workflows enabling faster debugging and issue resolution
  - Fixed GitHub rate limiting for customers with 10+ OpenAPI/AsyncAPI specs by switching from individual file fetching to repository cloning
  - Improved assistant reliability with backup LLM support, enhanced rate limit error handling, and more robust search tool functionality
 </Update>

 <Update label="August 16, 2025" tags={["Improvements"]} rss={{ title: "Weekly Updates", description: "Performance optimizations, LLM integrations, web editor polish, and API playground enhancements" }}>
@@ -1376,9 +1380,9 @@

  - OpenAPI to MCP conversion moved to backend, enabling hosted MCP servers to contain tools (expect clearer documentation and config options soon)
  - Added Ruby support to API playground
  - We added a feature such that you can now [specify API pages using just your docs.json](/api-playground/openapi-setup#auto-populate-api-pages) without creating any new mdx files.
  - Support for [`webhook pages`](/api-playground/openapi-setup#webhooks) in docs navigation from OpenAPI specs
  - Optimized AI model context by removing anchor link specifications from markdown links when navigating to Anthropic, OpenAI, or other providers

  ## Web editor improvements

@@ -1423,7 +1427,7 @@

  - Group-level public access: make entire page groups public via `docs.json` so you don’t need `public: true` on each page ([learn more](https://mintlify.com/docs/authentication-personalization/authentication-setup#group-level))
  - Support [`logoutURL in OAuth configuration`](https://mintlify.com/docs/authentication-personalization/authentication-setup#implementation-3) to delete upstream cookies and complete sign-out
  - On OAuth errors, users are redirected to your specified `logoutURL` to restart the auth flow
  - Fixed a flash of a 500 error during OAuth/JWT flows before the callback
  - Auto-strip `https://` from URLs in OAuth/JWT auth configuration to prevent misconfiguration

@@ -1441,9 +1445,9 @@

  ## Performance and reliability

  - All image and video assets present in your docs repo are now served at the appropriate path on your domain. For example, if you have `/assets/marketing/my-logo.png` in your repo, it will be available at `https://mydocsurl.extension/assets/marketing/my-logo.png`.
  - Email field on login for the Mintlify dashboard now autofocuses so you can start typing immediately _(quality of life improvement)_
  - Both custom domains and subdomains in Redis for a performance improvement on navigation load times (~50ms latency reduction)
  - Added retry logic for PDF exports to improve reliability
  - Fixed cookie consent popup reappearing after acceptance or dismissal—first selection is now respected
  - Fixed copying a page to clipboard on Safari by specifying a MIME `type` in `navigator.write`
@@ -1453,8 +1457,8 @@
  - CLI bugfixes for windows and pnpm, plus CI tests to prevent regressions
  - Improved error logging output—a quality of life upgrade for our engineering team when debugging
  - Minor fixes to the broken-link CI action when `contentDirectory` files are missing
  - Fixed a regression caused by the auth-protected preview fixes from last week where the active tab was not being set correctly in the navigation UI
  - Fixed theme light background color not being applied to active tab icons
  - Fixed an issue where changing the auth type in the dashboard would update and then flip back to the previously saved type—now the new selection persists after saving
  - Internal DX improvements for enterprise customers with custom UI libraries—it's now easier for us to include your components and accommodate requests on shorter timelines
 </Update>
@@ -1469,7 +1473,7 @@

  ## API playground enhancements

  - Fixed `multipart/form-data` file upload functionality in the API playground
  - Fixed anchor link behavior so clicking them updates the URL without scrolling to top of page
  - Fixed anchor link issues in nested tabs

@@ -1484,7 +1488,7 @@

  - Made search feel more crisp and accurate by aborting debounced requests as you type
  - Resource provisions for a new CDN - expect image asset and page load times to improve soon
  - Fixed bugs for rendering complex Mermaid diagrams like GANTT charts
  - Fixed CLI bugs on Windows to improve stability and added tests to prevent regression

  ## Technical improvements
@@ -1527,7 +1531,7 @@

  Learn more in the [assistant docs](/assistant/index).

  ## Subscribable changelogs

  - Automatically generate an RSS feed from changelog pages
  - Integrate RSS-enabled updates with Slack, email, and other tools
@@ -1551,10 +1555,10 @@
 </Update>

 <Update label="April 2025" tags={["New releases", "Improvements"]}>
  ## Web Editor 3.0

  <Frame>
    ![Webeditor3 Jpe](/images/webeditor3.jpeg)
  </Frame>
  Overhauled usability in the WYSIWYG editor.

@@ -1570,14 +1574,14 @@

  - Fixed top margin for changelog components
  - Improved reliability of right click behavior
  - After clicking publish, you’ll stay on the same page instead of being brought to an empty state
  - Standardized colors in file icons
  - Improved reliability after selecting new branches several times in a row
  - Removed Diff mode
  - More consistency when creating a new folder from the dropdown
  - Fixed block quotes creating more block quotes when trying to deselect

  ## AI Translations in beta

  <Frame>
    ![AI Translations graphic](/images/changelog/translations.png)
@@ -1611,7 +1615,7 @@

  - OG images fixed
  - Fixed icon style inconsistency for anchors without container
  - Improved styling nits for dashboard border for mobile-tablet-desktop responsiveness
  - Show code examples even when in simple mode for API playground
  - Support "command + k" shortcut for search in web editor
  - Codeblocks within callouts expand to fill the width of the callout area
@@ -1621,7 +1625,7 @@
  ## New configuration schema `docs.json`

  <Frame>
    ![docs.json screenshot](/images/changelog/docsjson.png)
  </Frame>
  We've introduced a new `docs.json` schema as a replacement for `mint.json`, to support better multi-level versioning, easier visual comprehension, and more consistent terminology. For more information on what's changed, [check out our blog](https://mintlify.com/blog/refactoring-mint-json-into-docs-json).

@@ -1641,7 +1645,7 @@

  1. Delete your old `mint.json` file and push your changes

  ## CI Checks

  Automatically lint your docs to find broken links, discover spelling and grammar issues, or enforce writing styles with your own Vale config. Learn more in our [docs](deploy/ci).

@@ -1649,7 +1653,7 @@

  All documentation pages are now automatically available as plain Markdown files—just append `.md` to the URL. This makes it easier for LLMs to ingest individual pages from your documentation.

  ## More Themes

  <Frame>
    ![graphic with text "Themes v2"](/images/changelog/themes.png)
@@ -1666,11 +1670,11 @@

  - [Guide to Technical Writing:](https://mintlify.com/guides/introduction)Best practices for writing technical documentation, including audience research, content types, and writing tips.
  - [Dropdown component](organize/navigation#dropdowns): Organize navigation with a dropdown, in addition to tabs and anchors.
  - [AI syntax fixer](https://x.com/ricardonunez_io/status/1892334887644123192): The web editor will catch if there’s a parsing error and use AI to suggest fixes.
 </Update>

 <Update label="January 2025" tags={["Improvements"]}>
  ## AI Assistant Improvements

  - New UI with dedicated chat page & pre-filled prompts
  - Stability improvements. For example, bug fixes of editing the wrong file or no files at all
@@ -1681,7 +1685,7 @@

  Customize access to any page or section of content depending on user permissions. Supports connecting with your own authentication system.

  ## Revamped API Playground

  We’ve overhauled the design and performance of the [API Playground](/api-playground/). Updates include:

@@ -1704,18 +1708,18 @@
 </Update>

 <Update label="November 2024">
  ## AI Writer

  <Frame>
    ![AI Assistant](/images/changelog/ai-assistant.jpg)
  </Frame>
  You can now ask AI to make changes to your docs, with the context of all existing documentation. Type in a prompt and the writer will propose changes by generating a pull request.

  ## GitLab Integration Upgrade

  We've improved our support for syncing with GitLab, such as enabling automated updates and preview deployments. Check out our [docs on GitLab](/deploy/gitlab) to get started.

  ## Web Editor

  <Frame>
    ![Web Editor](/images/changelog/webeditor.jpg)
@@ -1724,16 +1728,16 @@

  Check out our [docs on getting started with Web Editor](/editor).

  ## /llms.txt support

  <Frame>
    ![llms.txt support](/images/changelog/llms.jpg)
  </Frame>
  All docs instances are now automatically hosted at /llms.txt and /llms-full.txt so that LLMs can easily ingest your documentation. For more information, read the [docs on the new llms.txt standard.](https://llmstxt.org)

  ## Localization

  You can now localize your docs which operates similarly to versioning. Add a `locale` to a version and fixed content in Mintlify like "Was this page helpful?" will also match the locale.

  ### Quality Improvements

@@ -1749,14 +1753,14 @@
  <Frame>
    ![Changelog](/images/changelog/changelog.jpg)
  </Frame>
  ## Code Line Highlighting

  You can now highlight lines of code in your docs to emphasize and bring attention to important parts by adding a special comment after the language identifier. Use curly braces `{}` and specify line numbers or ranges separated by commas.

  ```javascript Line Highlighting Example {1,3,4,5}
  const greeting = "Hello, World!";
  function sayHello() {
    console.log(greeting);
  }
  sayHello();
  ```

diff --git a/es/assistant/configure.mdx b/es/assistant/configure.mdx
@@ -95,6 +95,17 @@ El assistant usa paquetes con asignaciones mensuales de mensajes. Un mensaje es
 
 Por defecto, el assistant permite excedentes. Puedes deshabilitar los excedentes para evitar incurrir en costos adicionales por uso más allá de tu paquete. Si alcanzas tu asignación de mensajes con los excedentes deshabilitados, el assistant no estará disponible hasta que tu asignación de mensajes se renueve. Si permites excedentes, cada mensaje más allá de tu asignación incurre en un cargo por excedente, pero los excedentes ocasionales pueden ser más baratos que subir a un paquete superior, dependiendo de tu uso.
 
+<div id="view-usage-breakdown">
+  ### Ver el desglose de uso
+</div>
+
+La página [Uso](https://dashboard.mintlify.com/settings/organization/usage) desglosa el consumo de créditos entre tus funciones de IA para que veas en qué se gastan los créditos e investigues picos.
+
+- Gráfico **Daily consumption**: un gráfico de barras apiladas del uso diario de créditos desglosado por producto (Assistant, Agent, Workflows, MCP y Others). Los totales por producto aparecen en la leyenda debajo del gráfico, junto con su participación en el uso total de créditos.
+- **Rango de fechas**: alterna entre **Last 7 days**, **Last 30 days**, **Last 90 days** y **Last 12 months** para ajustar la ventana de tiempo del gráfico.
+- **Filtro de producto**: elige **All products** para comparar todos los productos o selecciona uno (Assistant, Agent, Workflows, MCP o Others) para enfocar el gráfico solo en ese producto.
+- **Activity log**: una lista cronológica de los eventos que consumen créditos en todas las funciones de IA. Cada fila muestra la marca de tiempo, el producto y la fuente (por ejemplo, Web chat, Slack agent o Workflow) y los créditos utilizados. Selecciona **Load more** para ver eventos anteriores.
+
 <div id="change-your-assistant-tier">
   ### Cambiar tu paquete de assistant
 </div>

diff --git a/fr/assistant/configure.mdx b/fr/assistant/configure.mdx
@@ -95,6 +95,17 @@ L'assistant utilise des packages avec des allocations mensuelles de messages. Un
 
 Par défaut, l'assistant autorise les dépassements. Vous pouvez désactiver les dépassements pour éviter des coûts supplémentaires liés à une utilisation au-delà de votre package. Si vous atteignez votre allocation de messages avec les dépassements désactivés, l'assistant est indisponible jusqu'au renouvellement de votre allocation. Si vous autorisez les dépassements, chaque message au-delà de votre allocation entraîne un frais de dépassement, mais des dépassements occasionnels peuvent être moins chers que de passer au package supérieur selon votre utilisation.
 
+<div id="view-usage-breakdown">
+  ### Voir la répartition de l'utilisation
+</div>
+
+La page [Utilisation](https://dashboard.mintlify.com/settings/organization/usage) ventile la consommation de crédits entre vos fonctionnalités d'IA pour que vous puissiez voir où les crédits sont dépensés et enquêter sur les pics d'utilisation.
+
+- Graphique **Daily consumption** : un graphique à barres empilées de l'utilisation quotidienne des crédits, ventilée par produit (Assistant, Agent, Workflows, MCP et Others). Les totaux pour chaque produit apparaissent dans la légende sous le graphique, avec leur part de l'utilisation totale de crédits.
+- **Plage de dates** : choisissez entre **Last 7 days**, **Last 30 days**, **Last 90 days** et **Last 12 months** pour ajuster la fenêtre temporelle du graphique.
+- **Filtre par produit** : sélectionnez **All products** pour comparer tous les produits, ou choisissez un seul produit (Assistant, Agent, Workflows, MCP ou Others) pour concentrer le graphique sur ce produit uniquement.
+- **Activity log** : une liste chronologique des événements qui consomment des crédits dans toutes les fonctionnalités d'IA. Chaque ligne affiche l'horodatage, le produit et la source (par exemple, Web chat, Slack agent ou Workflow), et les crédits utilisés. Sélectionnez **Load more** pour parcourir les événements plus anciens.
+
 <div id="change-your-assistant-tier">
   ### Changer votre package d'assistant
 </div>

diff --git a/zh/assistant/configure.mdx b/zh/assistant/configure.mdx
@@ -95,6 +95,17 @@ AI 助手使用套餐，每月按消息配额计算。一条消息指用户与 A
 
 默认情况下，AI 助手允许超额使用。你可以禁用超额以避免超出套餐的额外费用。如果禁用超额后达到消息配额上限，AI 助手将不可用，直到消息配额重置。如果允许超额，超出配额的每条消息会产生超额费用，但根据你的使用情况，偶尔的超额可能比升级到更高套餐更便宜。
 
+<div id="view-usage-breakdown">
+  ### 查看用量明细
+</div>
+
+仪表盘的[用量](https://dashboard.mintlify.com/settings/organization/usage)页面会按 AI 功能分解你的积分消耗，便于你了解积分花在了哪里，并排查使用高峰。
+
+- **Daily consumption** 图表：按产品（Assistant、Agent、Workflows、MCP 和 Others）分类的每日积分使用量堆叠柱状图。各产品的总计显示在图表下方的图例中，并附上各自占总积分使用量的百分比。
+- **日期范围**：可在 **Last 7 days**、**Last 30 days**、**Last 90 days** 和 **Last 12 months** 之间切换，以调整图表的时间窗口。
+- **产品筛选**：选择 **All products** 比较所有产品，或选择单个产品（Assistant、Agent、Workflows、MCP 或 Others）以仅聚焦该产品。
+- **Activity log**：所有 AI 功能消耗积分的事件按时间排列的列表。每行显示时间戳、产品和来源（例如 Web chat、Slack agent 或 Workflow）以及消耗的积分数。选择 **Load more** 加载更早的事件。
+
 <div id="change-your-assistant-tier">
   ### 更改 AI 助手套餐
 </div>