Release notes

​

Highlights

​

New Clark features

• Submit feedback directly from the chat. Use the thumbs up or thumbs down controls to tell the Superblocks team how Clark is doing, without leaving your conversation. Your feedback goes straight to our team and helps us improve Clark’s responses.
• Automatically collapse long user messages. Lengthy messages, including detailed policy findings sent from the Publish tab’s Fix action, are condensed so they no longer push the rest of the conversation out of view.

​

Clark context management and reliability

• More context-efficient screenshotting. By default, Clark takes a small glance at the visible part of your app instead of rendering the whole page at full size. That uses about 7x fewer vision tokens for routine checks. When it needs more, Clark can capture a specific region, the full page, or a high-detail view.
• More context-efficient docs for Clark. We trimmed the docs Clark reads so lookups cost less. That leaves more room for your app in each prompt, uses fewer GAUs per prompt, and means less summarization.
• Clark knows it cannot capture certain images in screenshots. External images that block cross-origin requests appear with a clear placeholder, labeled “image not captured” on larger images, instead of a blank or loading state. Clark also gets a short note explaining the limitation, so it no longer retries the screenshot or assumes the image failed to render. That means faster generations and fewer GAUs used per prompt.
• Editor no longer shows a false “404 - Page not found”. Some apps generated by Clark would display a “Page not found” overlay in the editor, even though they ran fine in preview and when deployed. The editor now correctly detects these pages, so you no longer need to rework your app’s routing to clear the error.

​

Integrations

• APIs that download files now run faster. Binary REST responses are now sent in a much more compact format, so file download payloads are significantly smaller. This improves performance for any API that downloads files. For customers self-hosting data planes, upgrade your data plane to v1.53.0 or later.
• Upload binary data to REST APIs. Clark can now set the bodyType option to "raw", "binary", "formUrlencoded", or "multipartForm" to upload files, send form-encoded data, or post multipart forms to REST integrations in APIs. For customers self-hosting data planes, upgrade your data plane to v1.52.0 or later.
• Zendesk integration now supports OAuth 2.0 Authorization Code authentication. New Zendesk integrations default to OAuth, and you can use either a shared access token or per-user authentication. Each organization registers its own OAuth client in Zendesk admin, since Zendesk does not offer a shared one. Zendesk is phasing out API tokens, so switching to OAuth keeps your integration working. Existing integrations continue to use basic authentication until you switch them.
• OAuth2 Client Credentials integrations can now send credentials in the request body. A new Client authentication dropdown lets you choose between sending credentials as a Basic Auth header or in the body of the token request. This makes the Client Credentials flow compatible with APIs that only accept credentials in the body. For customers self-hosting data planes, upgrade your data plane to v1.52.0 or later.

​

Code Security

• Clark can now fix large batches of security findings. You can now fix a long list of security findings with Clark in a single turn. Clark breaks the findings into a checklist and works through the list, so you can address security issues before publishing.

​

Publishing

• You can roll back to a previously published commit even when pre-publish Policy Agents are enabled. Commits that already passed your pre-publish Policy Agents can be republished right away, without re-running the checks. This makes it faster to recover from production issues. To publish a commit that has not cleared the checks, reset the editor and go through the full policy review.

​

Editor UI

• The app preview banner has a fresh look. The old blue banner is now a neutral, dark design that matches the app builder, with a refreshed environment tag and a redesigned Publish button.

​

New Clark features

• Replace Excel-based workflows with production-grade apps. You can now upload an .xlsx file in the Clark editor or homepage chat, and Clark reads each sheet to understand its tables, headers, computed columns, and cross-sheet references. It then imports the sheets that hold data. Files can be up to 5 MiB.
• Clark can now import CSV data into your app’s database. Upload a CSV file in chat and Clark can load the contents into any connected data source available in Edit mode. This makes it straightforward to seed your development database from a spreadsheet or Excel workbook. Clark can also process your CSV and send the data to any connected integration.
• Checkpoint cards in the Clark panel are cleaner and easier to act on. The newest and live cards keep their Publish and Preview buttons, and every card keeps a labeled Preview so you can jump back to any version. On older cards, Restore and publish details appear when you hover, and a publish in progress shows as progress you can cancel instead of a grayed-out button.

​

Clark context management and reliability

• Clark now caps oversized tool error messages. Errors from validation, blocked access, missing tools, and confirmation-required steps are now limited to 25,000 characters, matching other tool failures. This keeps Clark’s context lean, while short errors and permission behavior stay the same.
• Clark is more resilient to brief editor connection drops during generation. If your editor unexpectedly loses its connection while Clark is building, Clark’s next browser actions, like navigating the preview or taking screenshots, now wait up to 30 seconds for the editor to reconnect instead of failing right away.
• The Clear chat history button is now disabled while Clark is generating. This prevents history from being cleared mid-turn. Stop remains available, and clearing is re-enabled as soon as the turn stops.

​

Integrations

• You can now download files from REST API integrations. Pull a generated PDF report, an invoice, or an image from a REST API and use it in your app or pass it along to another integration. For customers self-hosting data planes, upgrade your data plane to v1.50.0 or later.

​

Billing and usage

• You can now filter GAU usage on the Billing page by activity type. Filter GAUs to Clark usage or Code Security activity (Security Fixes, Security Agents, Security Scans, and CVE Scans), so you can see how much your organization spends securing applications.

​

Code Security

• Policy Agents no longer run for Git-connected apps. Pre-publish Policy Agents are now skipped for apps connected to Git. Run security and code checks in your CI pipeline for those apps. If you disconnect Git from an app, any enabled policy agents apply on the next publish.

​

Clark Context Management & Reliability

• Clark now catches problematic attachments before sending your message. Clark rejects unsupported, oversized, or unreadable files before your message is saved or sent to the model. The error shows the filename, the limit for that file type, and how to fix it, and your draft and valid files are kept so you can retry.
• Clark sessions are faster and cheaper with improved prompt caching. Clark no longer rewrites messages it has already sent each time it captures a screenshot, so prompt-cache entries stay valid and long build sessions cost less per prompt. Screenshot resolution has also been adjusted, and the stored session record is now size-bounded.
• Clark chat updates now stay in sync across open editor tabs. Clark’s responses and progress updates now appear in every editor tab you have open for the same app, even if another tab is briefly reconnecting. This fixes a case where a second tab could show “Thinking…” until you reloaded.
• Clark recovers from stuck chats caused by malformed tool calls. Clark now repairs malformed tool names in saved chat history before sending it to the model. Retries no longer get stuck on a rejected conversation, so you don’t need to clear chat history to continue building.
• Clark re-reads files less after making edits. Clark now avoids reading a file again right after it edits or writes it, since a reported success means the change landed. This cuts unnecessary steps and context usage, so editing sessions stay more efficient.

​

Billing & Usage

• Policy Agent runs are now attributed to users. AI usage from Policy Agents is now attributed to the person who started the run. It appears under that user on the Usage page.
• Usage page now shows security and remediation GAU costs per checkpoint. See exactly how much you spend to secure each checkpoint. Checkpoints with Policy Agents now expand into Clark Usage, Security Agent, Security Scan, and CVE Scan totals, covering both the cost of each review and the cost of addressing its findings.
• Usage page now shows all your usage, including before a contract change. The Usage page now displays usage in units only, with dollar amounts shown only in the commitment bar. Usage recorded before a contract change, such as a trial-to-POC upgrade, now appears, and the Published app charges and Governed Agents Usage tiles are consistent across plans.

​

Code Security

• New Policy Agent flags vulnerable dependencies before you publish. A new pre-publish Policy Agent for Vulnerability Detection checks your app’s dependencies against known CVEs before publishing completes, preventing your builders from shipping apps with known CVEs. Clark will upgrade or find an alternative package that’s safe for production.
• Security findings stay consistent between publishes. The policy gate and Security Agent now report consistent findings across republishes when your code hasn’t changed. Issues no longer disappear, show up as duplicates, or change unexpectedly, making it easier to find, fix, and publish.
• Dependency scan findings now show the vulnerability name. Findings from the publish-time dependency scan now show the affected package and version alongside the advisory name, making them easier to triage.
• Security emails now link straight to Fix with Clark. Vulnerability emails (new finding alerts and the admin and creator daily digests) now take you directly to Fix with Clark for the affected app and finding, matching in-app notifications. The emails also have a cleaner layout, with color-coded severity badges and useful links.

​

App Publishing

• Clark publish button no longer times out during slower builds. The publish button in Clark chat now waits up to 15 minutes, up from 5, before showing “Timed out — Retry”. Slower builds, such as those that start after an idle period, can finish without a false timeout message.
• Publish tab no longer shows a false “older version” warning on your latest version. The Publish tab now treats a checkpoint and the commit created from it as the same version. Opening your latest version from history no longer triggers a restore-version warning, and the header and pipeline stay consistent.
• Publish tab now explains when you lack publish permission. There is now a clear message on the Publish tab when you don’t have permission to publish the app, with guidance to contact an admin. You can still create a Preview if you have permission.
• The Live tag in the Publish tab now opens your published app. Clicking Live in the Publish tab header of the 2.0 app editor now opens the deployed app in a new tab. This matches how the Live badge already works in chat checkpoint cards and the history panel.

​

Embedded Apps

• Embedded previews stay working after you publish a new version. Embedded app previews that were already open no longer fail with “not found” errors when a new version is published. Open sessions keep loading the version they started on, so users aren’t interrupted mid-session.

​

Environments & Data Tags

• Switch environments for your editor session without changing app settings. You and Clark can now change the selected environment (data tag) for your current session without touching saved app settings. Use the new “Selected for this session” picker in the Edit section of the Environments sidebar, or ask Clark, who will confirm with you before switching.
• New permission to manage app environments. A new permission determines who can add or remove environments in an app. This enables admins to keep sensitive environments, like a restricted Dev or Prod variant, out of builders’ apps unless they are explicitly added by a user with this permission. The built-in Developer role includes this permission, but you can create a custom role that excludes this permission.

​

New Clark features

• Clark can now run approved schema changes on external PostgreSQL databases. In Build mode, Clark can run a single DDL statement, such as creating or altering tables, indexes, and views, against external PostgreSQL-family integrations after you approve it.
• Clark can now uninstall packages from your apps. Clark can remove packages as well as add them, which makes it possible to fix vulnerabilities that require dropping a dependency.
• Clark chat history now appears right away when you open the editor. Recent messages now show up as soon as the editor opens, without waiting for the live-edit environment to connect. You can also scroll back to load older messages in long conversations, so earlier history stays available.

​

Clark context management and reliability

• Clark reads files more efficiently with fewer wasted tokens. When Clark reads specific line ranges of a file, the reported range now matches the lines actually returned. Overlapping ranges are merged and returned in file order, so no lines are repeated. This cuts token usage during builds.
• Clark no longer fails with a “Maximum call stack size exceeded” error during error logging. Clark now handles error logging reliably, so a logging problem no longer surfaces as “Failed to get response from LLM provider”.
• Clark handles large API test errors without losing context. When an SDK API test fails with a very large validation error, Clark now shortens the error before reading it, so it no longer fills the model’s context. The actionable message is kept first, and small errors are shown unchanged.
• Clark no longer fails on conversations containing emojis. Long Clark conversations with emojis or other special characters now compact cleanly, so they no longer cause model errors. Existing conversations that were affected can continue without clearing history.
• Invalid PDF attachments are now caught at upload with a clear error. Files that look like PDFs but are not valid are rejected when you upload them, so they no longer break the rest of your chat. If a model provider rejects an attachment, Clark now explains the cause and what to do next instead of showing a generic error that suggests retrying.
• Switching back to Clark Chat no longer flashes a loading screen. Returning to Chat from the Versions or Environments tabs in the Clark panel now shows your conversation right away, so you no longer see a brief loading skeleton or flicker.

​

Integrations and APIs

• API code can now read the current data tag with ctx.dataTag. APIs can read the server-validated data tag (such as staging or production) via ctx.dataTag, so you can run checks like permission assertions only against production data. For customers self-hosting data planes, upgrade your data plane to v1.51.0 or later.
• APIs can now scan DynamoDB tables larger than 1 MB. DynamoDB scans in your APIs can now page through large tables instead of silently returning a partial result. New options let you pass a start key, index name, limit, and related AWS fields. Existing scan and table query calls work unchanged. For customers self-hosting data planes, upgrade your data plane to v1.51.0 or later.
• Test Connection now works for secrets stores attached to multiple environments. Testing a secrets store connection no longer fails with a 401 error when the store is attached to multiple data tags and uses named agents. Test Connection now uses the same profile your agent routing selects, so you see real connection results instead of profile errors.

​

Editor and UI polish

• Homepage now shows up to 20 recent apps in a scrollable grid. The recent apps section expands from 3 apps to a wrapping grid of up to 20, about five per row, so you can jump back into your work faster.
• Refreshed look for the Integrations and Apps pages. The Integrations and Apps pages have a cleaner light-theme design. Integration forms disable Save, Cancel, and Delete while a connection test runs, and the Apps page shows folders and apps together in one grid. Integrations now load with a skeleton instead of flashing an empty state.

​

Environments and data tags

• Environments and self-hosted data planes now live on one Data setup page. Admins can now configure data tags and the data planes that serve them in one place. Clicking a tag highlights the data planes that serve it, and each tag shows how many active data planes cover it.
• The data plane setup wizard is now a paged flow. The wizard uses Next and Back instead of one long scrolling page. When App Databases is enabled, it collects data-plane identity and AWS networking inputs and generates Terraform and Helm configs from them.
• You can now restore a removed Default integration configuration. If you have removed the Default configuration for an integration, the options to add Default and tagged configurations now appear again. Restoring Default opens a fresh form ready for editing, and creating tagged or cloned configurations works as before.

​

Highlights

• Admins can now set a default data tag per environment. This data tag will be the only tag available by default in all new apps.
• We’ve introduced a new permission for apps:manage_environments. This permission controls whether a builder can modify the data tags available to their app, so admins can prevent builders from enabling access to sensitive integrations configurations and data planes.
• Clark now has information about the available data tags per environment, plus the ability to add and remove data tags from an app for builders with this permission.
• All of these enhancements are important for App Databases, since the available data tags determine the database lifecycle.

​

Clark AI

• Increased context window size: Clark now has 2x the context window, reducing summarization and improving Clark’s focus in longer sessions.
• Improved context management for Clark reading API documentation. Clark no longer reads full API documentation files, reducing context bloat and summarization.

​

Code Security

• Added the ability to “Fix with Clark” for any CVE from the Security Center.

​

Everything else

• Clark will no longer incorrectly guide users to use Git to publish their applications for any apps that are not connected to Git repositories.
• Show lock screen instead of crash when another user is currently eding an application
• Fixed a misleading “Token took too long to load” error shown to users viewing published apps
• Redesigned edit mode error empty state when the editor fails to load
• Fixed bug where “Clear chat” only cleared the session history for the current user. Clearing your chat history now clears the history for all builders as expected.
• Any user with billing:manage permission can now see the checkpoint labels for all checkpoints in the Billing & Usage page. Previously the label was incorrectly redacted unless the viewer had the org-wide apps:update permission.

​

General improvements

• Idle users no longer prevent other users from editing apps. If a user steps away or their computer goes to sleep, teammates can now take over their editing session instead of waiting for the editor to explicitly close their editing window.
• Upload images to Clark from the Knowledge page: Fixed bugs with images uploaded through the Knowledge Clark chat that were resulting in errors.

​

Code Security

• **Notifications for new CVEs: Every hour, Superblocks scans all packages across your published apps for new CVEs and notifies you when something new appears. You’ll receive an hourly in-platform notification in your inbox, plus a daily email digest.
• ** You can now drill into the App detail page from the Security Center to view the software bill of materials (SBOM) for the currently published version and all historical versions.
• Advisory and Blocking with Exceptions: You can now set Security Scans and Security Agents to run as Advisory or Blocking with Exceptions. Advisory Policy Agents do not block publishing. Blocking with Exceptions prevents publishing, but builders can bypass and Publish anyway for false positives.

​

Integrations Platform

• REST integrations support OAuth with PKCE. Authorization-code flows can use PKCE for both public and confidential clients. If your organization is self-hosting your data plane, you’ll need to upgrade to v1.49.0 to access this functionality.
• Admins can now delete the default integration configuration. This is relevant if an integration is not accessible for certain data tags; for example, if an integration is only accessible from a specific on-premise agent due to network isolation.
• “Connect your data” is hidden from the Clark chat for users that don’t have permission to create integrations.

​

Clark AI

• Fewer false write-approval prompts on SQL. Clark ignores -- comments when deciding whether an API is read-only, so words in comments (for example “set” in “set of”) no longer trigger an approval prompt for a read.
• Improved stability for app upgrades. Legacy apps with hundreds of APIs can be upgraded without running out of memory.

​

Everything else

• A new publish does not break the app for any live users. Open sessions keep running on the version that a user has loaded instead of breaking when a new version goes live. The previous version stays available for up to 12 hours after it’s replaced.
• Smoother app loading. Opening an app now shows one continuous progress bar through startup instead of a spinner that restarts at each step.
• Long “Created by” names (such as email addresses) no longer overflow the homepage list.

​

Highlights

• Every checkpoint in the Clark chat can be built into a Preview or published to Production
• You no longer need to manually commit just to generate a Preview
• Version history lives directly in the Clark chat panel — click the versions icon at the top of chat
• Every Preview and Publish has a detail view for progress and errors, and is where pre-publish actions such as security scans and database operations are surfaced

• GAU usage rolls up to the relevant Clark checkpoint, so you can see what it cost to deliver a specific outcome
• Group usage by User, App, and Day to find the highest-ROI work
• See which published apps are charged each month, including pending charges
• Break spend down over time across GAUs, published apps, users, and more with charts and filters

​

Clark AI

• Clark can write to integrations with your approval. In Build Mode, Clark can run one-off data updates and similar mutating API or database calls from chat after you approve each action.
• More accurate read vs write detection for Postgres. Clark uses a real PostgreSQL parser instead of keyword matching, so you get fewer false permission prompts and fewer accidental write allowances.
• Better Clark context management. Clark does less redundant summarization mid-conversation, so it stays smarter and more consistent throughout longer sessions.
• Clark context is shared for the whole app. Clark’s reasoning context is scoped to the application, not each individual editor. Previously, the chat could look shared while Clark was missing another teammate’s context.
• Clark no longer gets stuck on “Thinking…”. If the connection drops mid-turn, Clark recovers and continues instead of leaving you hanging with no response.
• Clark @-mention category drill-down. Clicking a category such as APIs or a database lists the entities in that category instead of showing “No entities available.”
• Chat messages no longer scroll from the top when loading. Opening a conversation now lands directly on your newest message, instead of visibly sweeping down through the history while it loads.

​

Everything else

• Filter homepage apps by deployment status. The apps list supports All / Deployed / Not deployed. Unpublished apps show “Not deployed” instead of a stale last-published date.
• Self-serve teams are no longer limited by builder seats. Self-serve / Teams plans no longer use seat-based pricing, so your whole team can use Clark without being constrained by the number of builder seats.
• Upload images up to 50MB. Give Clark larger screenshots and reference images as visual context for building your apps.
• Rotate Git tokens without disconnecting. When Git Setup is healthy, you can update your Git credential in place instead of disconnecting and reconnecting.
• Shareable published app URLs stay clean. Published app links no longer append an internal version query param, so URLs you copy and share remain stable.
• Admins can inventory packages across the org. Listing distinct npm packages used across your organization (for security and supply-chain reviews) now requires the policies:manage permission, which Admins and Owners have by default. You can also assign that permission to other roles.
• IdP re-auth in published apps. When an On-Behalf-Of OAuth token expires, the re-auth popup opens from a real user gesture instead of leaving end users in a dead-end state where APIs cannot execute.
• SCIM display names. User updates now keep the display name sent by your identity provider.

• Clark API testing stays scoped. When you edit an API in a shared file, Clark tests only that API instead of forcing tests on unrelated APIs in the same entry point.
• Typed answers to multiple-choice questions appear in order. If you type a custom answer instead of clicking a choice, it shows up in chat in the right place before Clark continues.
• Clarifying text during tool-permission prompts is guidance. Extra text you enter while approving or denying a tool call is passed to Clark as guidance instead of being treated as a denial.
• Clark chat no longer flashes in the side panel. On initial load, chat settles in the center without first appearing briefly in the side panel.

• No Superblocks top bar on published apps. Published apps no longer show the black bar with built-in Superblocks branding.
• Publish button is only visible for users with permission to deploy. Builders without deploy access no longer see a Publish button they cannot use.
• Homepage visual polish. Refreshed the homepage color palette, icons, and avatars for a cleaner look and feel.
• Non-JSON API responses. Superblocks can now connect to API endpoints that return XML, CSV, HTML, or plain text. Previously, responses were decoded as JSON and failed on anything else; you can now request a text response and get the raw body back.
• Integration forms show saved bound field values on first paint. Bound fields no longer appear blank until you focus them.

​

Highlights

​

Everything else

• Clark now maintains a checklist to track and manage tasks during large builds and migrations, making it more accurate and effective for complex, multi-step work
• Clark can now navigate between pages to take screenshots, enabling it to verify multi-page apps and catch rendering issues across routes
• Clark now self-fixes common import errors (e.g., type-only imports used as values) that would previously cause blank previews and block UI verification
• Better tool timeout handling — long-running tools like API verification now have appropriate timeouts (up to 5 minutes), reducing false failures on slow API or LLM calls

• Embedded app bootstrap is more reliable — the iframe handshake now retries on both sides so slow-loading embeds complete instead of timing out with “Authentication Error”
• Fixed data tag switching in embedded 2.0 apps — setDataTag() now correctly re-syncs the profile in the deployed shell
• Commit preview now works correctly for code-mode apps that have commits but have never been deployed

• Fixed a bug where workflows called from an app would use the wrong environment — the data tag display name was being sent instead of the data tag key
• Fixed Databricks and other OAuth integrations getting stuck for up to 10 minutes on an expired token — tokens are now reactively evicted and re-exchanged when a provider rejects them
• Fixed GraphQL and other integrations failing to save due to validation errors

• Fixed “Cannot commit in draft mode” errors that blocked users from committing even when they were not in an active AI session, caused by an uninitialized or stuck draft state
• Commit errors now show the real error message instead of [object Object] in the toast notification

• Streamlit is now a first-class app import source in the migration modal, with its own icon, prompt copy, and detection signals

• New MCP tool list_distinct_packages_in_org lets a security or platform owner enumerate the distinct npm package footprint across an entire org in a single call

​

Highlights

​

Everything else

• Increased Clark text file upload limit to 3MB for better support of larger documents and datasets
• Fixed out-of-order chat messages when typing custom prompts instead of using multiple choice widgets
• Eliminated thinking indicator jitter when knowledge messages arrive during conversations
• Added automatic and manual AI credit top-up for self-serve customers so you can build seamlessly without being interrupted — Learn more →
• Added geolocation API support for deployed applications — apps can now access navigator.geolocation when embedded

​

Clark AI Improvements

• Fixed issues where Clark would get stuck on “Summarizing conversation” after connection interruptions
• Improved the restore functionality in Clark’s checkpoint interface

​

Integrations & Authentication

• Enhanced the AI Builder license assignment flow for adding team members
• Added a Usage tab to integration pages showing which applications reference each integration
• Improved support for on-behalf-of token exchange integrations, allowing Clark to test APIs that require delegated authentication
• Fixed scrolling and multiple selection issues when choosing integrations in the chat interface

​

Editor & Development Experience

• Added Input/Output preview pills to API request panels with detailed popovers
• Added end-to-end timing summaries to SDK API trace views
• Users can now skip the datasource selection step in the onboarding survey
• Improved the integration home page filters with better “select all” and “deselect all” behavior
• Enhanced visual feedback when Clark is testing APIs and requesting permission to test write operations
• Various UI polish including improved chat loading animations

​

Billing & Usage Improvements

• Improved trial-ending emails with clearer billing information and direct links to billing management
• Removed unnecessary Stripe bundle loading from deployed applications for better performance
• Added audit events for application undeployment actions

​

Highlights

• Role Management — Assign and update organization-level roles (Admin, Developer, End-User) for users and groups from your favorite AI tool. Useful for onboarding automation, bulk role changes, and scripted access control.
• Chat History Retrieval — Retrieve the full conversation history between your builders and Clark AI for any application. See what prompts were sent, which plans were approved, and what screenshots were attached. Valuable for auditing how apps were built, onboarding new team members to an existing app’s context, examining the intent behind AI-generated code, and identifying potential abuse.
• Audit Event Aggregation — Run server-side analytics over your organization’s audit log without downloading raw events. Answer questions like “how many unique builders were active this month?”, “what’s our daily API execution volume?”, and “which apps have the most activity?” — makes it easy to report on your team’s usage and take action.

​

Everything else

• Clark sidebar now shows actual knowledge entry names instead of generic “Read app knowledge”
• Prevented users from uploading unsupported file types which would break Clark’s chat context and result in chat errors
• Fixed cmd+click and hover preview for links in deployed and preview apps
• Fixed integration setup forms not populating field values reliably when viewing or editing connections
• Added auto-focus on search input in the “Connect your data” modal
• Fixed Clark checking integration create permissions before opening setup forms — no more confusing permission errors after configuring the fields because you don’t have permission on the integration
• Improved Clark’s error handling when integration metadata isn’t available
• Added a grid view toggle to the apps homepage — switch between the familiar table view and a new visual grid with app screenshots for easier browsing

​

Highlights

1. A detailed trace viewer to understand individual queries and timings run as part of your API
2. Real-time visibility into API calls as you use your app, making it simpler to correlate your UI with specific APIs
3. A code editor in the browser for reviewing the underlying React and TypeScript code

1. All Cloud customers
2. OPA customers on v1.37.1 or higher

​

Clark improvements

• Added the ability to commit, build, and publish right from a checkpoint in the Clark chat
• Clark is now more effective at building custom logic based on user attributes
• Clark is now more effective at building custom logic using Data Tags
• Fixed display issues when referencing integrations with special characters in their names
• Chat now properly auto-scrolls to new messages, fixing issues where auto-scroll would break
• Clark no longer incorrectly requests permission to test read-only APIs
• Fixed “No metadata found” errors when Clark explores integration capabilities
• Copy buttons for code blocks: hover-to-reveal copy button for syntax-highlighted code blocks in Clark chat
• Expandable package install messages: “Installed X packages” messages can now be expanded to see which packages were installed

​

Integrations & authentication

• Eliminated “Authentication token scope value is incorrectly formatted” errors for Slack per-user OAuth
• Valid identity provider JWTs now refresh silently instead of opening new browser tabs
• Better integration sorting: page now sorts alphabetically by default with properly working “Created by” column sorting
• Fixed blank integrations screen for new organizations with no connected integrations

​

User management

• Cross-org user transfers now work: if someone on your team mistakenly creates their own organization, you can invite them to join your org and they’ll be transferred over

​

Billing and plans

• Fixed blank billing page for POC and other non-standard plans
• Clearer prorated costs: preview invoice for Teams customers now shows accurate prorated amounts instead of confusing totals

​

Navigation and editor

• “Start building” button in Clark introduction modal now navigates to homepage
• Fixed folder breadcrumbs showing org name instead of “Apps” when navigating folders
• Version panel now links to appropriate docs based on both template type and Git configuration
• Added helpful guidance when attempting to commit without Git configured
• Hidden Developer Preferences panel for fullstack apps where it’s not applicable
• Route load errors are now properly forwarded to Clark AI for diagnosis
• Fine-grained progress for editor start: live edit loading now shows a 4-step progress indicator tied to real connection phases instead of generic rotating messages
• Simplified sidebar: removed pin/unpin feature and added close buttons to all panels
• Fixed returning from deployed app going to /integrations instead of /home

​

Bug fixes

• Added safety measures to prevent file corruption during multi-file edits
• Integration names now appear correctly in Superblocks 2.0 API trace views
• OAuth setup errors now properly surface to users instead of failing silently
• Redesigned email templates: new clean card layout for org invites, transfers, verification, and share invites
• Fixed bulk delete on home page failing with transaction errors

​

Clark AI

• File uploads re-enabled across all Clark surfaces — app conversations, the homepage, and Knowledge
• Clark now references the Superblocks public docs, making it better at working with integrations and other platform-specific features
• Several improvements to Clark’s context management — Clark stays focused better throughout longer sessions, large files are automatically size-limited, and compaction no longer drops active task state
• Enabled SMTP plugin support for Clark — you can now build email-sending workflows with SMTP integrations
• Clark no longer disconnects mid-generation if you switch to another browser tab
• Fixed a bug where chat history was lost on page refresh during active generation
• Fixed Clark attempting to edit code while in plan mode — a regression that caused unnecessary correction loops

​

Clark Knowledge

• Empty state added to the Knowledge canvas when no item is selected
• Added a link from the in-app Knowledge panel to the admin Knowledge page for easier management
• Fixed knowledge promotion failing after a previous hotfix
• Fixed a permissions issue that blocked saving integration knowledge with a 403 error
• Friendlier error messages in Knowledge chat when issues occur

​

Billing & self-serve

• New AI Credit Usage chart with daily usage breakdown
• Teams plan users can self-serve purchase additional deployed apps

​

Integrations

• Fixed the save modal for integrations with data tag configurations — the connection test flow now works correctly when data tags are set
• Fixed creating integrations in Clark AI mode failing for certain plugins (e.g., Salesforce) due to a schema validation error

​

Source Control with Git (beta)

• Ask Clark to help you set up and manage your Git repository — connect to a repo, commit changes, view diffs, and check branch status from the chat
• AI-generated PR descriptions when opening GitHub PRs from Superblocks
• Clearer error messages when commits fail — including specific guidance for insufficient GitHub PAT permissions

​

Bug fixes

• Fixed BigQuery auth being stripped in deployed mode, causing token expiration for customers using Workforce Identity Federation
• Fixed Databricks Token Federation re-authentication — deployed apps now correctly prompt for re-auth, and the editor re-executes APIs after authenticating
• Fixed long links breaking the chat panel layout

​

Highlights

• Clark now accepts file uploads when building apps. You can share PDFs, CSVs, Markdown docs, and other files directly with Clark — both in app conversations and from the Superblocks homepage. Give Clark your actual schemas, data, or docs to work from instead of describing them by hand.
• Promote app knowledge to your whole organization. Insights you develop while building — about your integrations, data, or workflows — can now be promoted to broader scopes: org-wide, personal, or integration-level. Knowledge compounds across your team instead of staying locked to a single app.

​

Other improvements

• Tag integrations when starting a conversation from the homepage — Clark gets immediate context about your data sources before you even create an app

• Markdown tables now render correctly in Knowledge forms and the rich-text editor
• Knowledge page updated with inline actions, a refreshed icon, and cleaner empty-state copy

• New Usage & Credits page with role-based billing action
• Plans & Credits page updated with a builder seats table, seat assignment modal, and checkout flow for Teams customers

​

Bug fixes

• Fixed up to 3 duplicate API calls firing on page load when only 1 should run
• Fixed WebSocket and AI stream disconnects going undetected and added better observability to surface these failures proactively
• Minor fix to integration tagging height on the Clark homepage

​

Highlights

• File attachments in Clark Knowledge. Upload documents like design guideline PDFs, CSS stylesheets, internal docs, or security guidelines; Clark uses them to build out knowledge optimized for his own consumption when building. No more copying and pasting content in manually.
• Start building from superblocks.com. Type a prompt directly on the marketing site and land in a working app. You can start building even before you log in.

​

Bug fixes

• Fixed a bug where Clark could time out when testing APIs
• Improved Clark API generation accuracy by fixing an SDK documentation regression — Clark now generates working APIs more reliably on the first try, with fewer correction loops
• Fixed UI flickering that occurred when Clark reads system skills during prompting
• Fixed certain read-only APIs not correctly triggering approval prompts when Clark attempts to test them
• Fixed cleared Clark chats reappearing after a checkpoint refresh
• Fixed queued prompts causing horizontal scrolling in the AI chat sidebar
• Fixed an issue causing AI model errors for some users
• Fixed a handoff bug that could leave apps uneditable when switching between users

​

Highlights

• Clark is now available to everyone. Clark AI is out of private beta and open to all. New users can sign up for free at superblocks.com, start a free trial, and upgrade to the Teams plan entirely self-serve. No more waitlist.
• New Snowflake integrations: Cortex AI and Postgres. Superblocks now connects natively to both Snowflake Cortex AI and Snowflake Postgres. If your team is in the Snowflake ecosystem, you can bring AI-powered queries and Postgres-compatible data sources into your apps seamlessly.
• A new homepage for building. The new Clark homepage puts a prompt box front and center. The first thing you see when you log in is the Clark prompt box and your most recently edited apps, so you can start building immediately.

​

Onboarding

• New onboarding survey flow helps tailor the experience when you first sign up and start using Superblocks
• We added a banner to the apps list to help existing Low Code users easily navigate between Clark AI and their legacy apps and workflows

​

Clark improvements

• The Org Knowledge editing experience now includes an accept/reject diff workflow. You get an explicit review step before any changes are saved, and the Knowledge panel refreshes immediately after.

​

Bug fixes

• AI/Code Mode requests no longer fail when app screenshots are very large. They’re now scaled down automatically.
• Fixed a bug where saving an app knowledge fact could put the app into a broken state
• Fixed a bug where Clark would stop responding after you replied to a multiple-choice question about app knowledge
• Fixed a routing issue where Clark’s app view would be stuck in the center of the screen when routes were defined in a variable
• Prompts containing UUIDs are no longer incorrectly blocked by chat guardrails
• Downloaded app file names now reflect the actual app name
• Fixed the API panel getting stuck open with no way to exit in code mode
• Fixed a visual flicker and improved save UX in the Edit Data Tag modal
• Fixed a bug where the REST Base URL was silently discarded when saving an integration
• Fixed a package installation failure in cloud edit mode caused by npm being overridden by pnpm
• Default data tag selection is now pre-populated for a smoother setup experience

​

Highlights

• Clark in Organization Knowledge. You can now chat with Clark directly in the admin panel to set up your organization knowledge. Clark helps you articulate your org’s conventions, integration nuances, and personal preferences through conversation, making it faster and easier to get your knowledge dialed in. As a reminder, knowledge works at three levels that Clark follows by default:
• App Knowledge: Clark remembers context per app. You can now ask Clark to remember facts, architecture decisions, and context specific to an app — just prompt it during a session. Clark stores these as app-level skills and uses them in future sessions, so the more you build, the better Clark gets. No setup required — just tell Clark what to remember.
• Environment access controls. Admins can now configure which data tags are available in each Superblocks environment — edit, preview, and production — directly from the new Environments admin UI. This makes it easy to lock down production data so builders and Clark only work against the right environments.

​

What’s next

• Knowledge promotion. Soon, you’ll be able to save knowledge directly to your organization from chat while building apps. Today, you can save app knowledge during a build session, and org knowledge when chatting with Clark in the admin. Knowledge promotion bridges the gap, so insights you surface while building flow back to your org knowledge without leaving the app editor.
• Improved onboarding experience. We’re working on making it easier to get started with Superblocks, including a new home page, the ability to set up integrations by talking to Clark, and a more seamless commit and deploy flow.
• APIs as code. We’re moving to code-based APIs written entirely in TypeScript — a new format that Clark can generate more accurately and is easier for you and your agents to review and maintain.
• Source Control for Clark Apps. We’re adding the ability to sync your Clark app code to remote Git repositories for code review and automation.

​

Other improvements

• Web exploration links are now clickable in the results panel for easier access to referenced sources
• Open share modal immediately when you create an integration for easy access control management
• Restored manual JavaScript/Python/REST/GraphQL step creation in API editor
• Increased UI timeout for OPA healthchecks to reduce “unable to connect to on-premise agent” errors
• App debug downloads now use zip format for easy sharing

​

Bug fixes

• Fixed issue where Clark would hang indefinitely when testing APIs — the system now properly handles timeouts and error states
• Fixed bug where reference images uploaded with the first prompt were lost and couldn’t be included in plans
• Fixed bugs with API runs showing error exclamation points and green checkmarks on a single run

​

Highlights

• Queue prompts and steer Clark mid-generation. You can now send messages while Clark is working — queue up follow-up instructions as ideas come to you, or interrupt and redirect Clark mid-generation if it’s heading the wrong direction. No more waiting for Clark to finish before giving your next thought.
• Organization Knowledge beta is now open to all customers. We’ve opened up the Org Knowledge beta to everyone. You can define knowledge at three levels that Clark follows by default:
  • Organization — design system and brand guidelines, coding standards, team conventions, and other org-wide context
  • Integration — nuances about your data sources, schema quirks, naming conventions, query patterns, and other details that help Clark work with your APIs correctly
  • Personal — your individual preferences and how you’d like Clark to work with you

​

Clark improvements

• Screenshots now capture full pages and wait for content to fully load before analyzing, so Clark no longer reasons over incomplete UI states
• Improved context summarization to better track build completion — Clark no longer asks about next steps after successfully making proposed changes
• Added a warning when users enable too many Knowledge entries

​

Data exploration in Clark Chat

• Enhanced JSON view inside results tables for better readability
• Improved handling of truncated API outputs with clear messaging and partial result display

​

Bug fixes

• Fixed retry UI showing incorrect success messages when AI model requests were paused or failed
• Fixed success toasts incorrectly showing for failed API calls
• Clark no longer incorrectly requests permission to test read-only APIs
• Clearing context now properly clears in-memory data to prevent recurring inference errors in edge cases
• Fixed checkpoint restoration to avoid a rare race condition where users could not restore to past checkpoints

​

Highlights

• Clark is now powered by Claude Opus 4.6. Clark now runs on Anthropic’s Claude Opus 4.6 — the world’s best coding model. You should see improvements across the board in code generation quality, reasoning, and Clark’s ability to handle complex app-building tasks.
• Clark automatically optimizes for large datasets. Apps generated by Clark now automatically implement pagination, virtualization, and server-side aggregation when working with large datasets. This means fewer browser slowdowns and better performance out of the box — no extra configuration needed.
• Agent Memory is in Beta. We’re rolling out the first piece of Clark’s memory system: organization-level knowledge. Admins can now define facts about your org, integrations, and users that Clark follows by default — things like coding standards, integration quirks, and team conventions. Reach out for access!

​

Clark improvements

• Clark now handles API call failures gracefully — errors display in the UI instead of appearing to hang indefinitely
• You can now skip testing specific APIs and still complete generation when Clark encounters API test rejections
• Improved Clark’s ability to inspect historical API runs
• Fixed false positives in safety checks that were blocking legitimate internal security audits

​

Editor experience

• Added mobile view support on the editor canvas
• Fixed profile switcher being disabled when chat is in fullscreen mode
• Returning to the editor after idle time no longer shows a fullscreen connection error — it automatically reconnects
• Fixed chat sidebar opening before the app has fully loaded
• Fixed pages opening in new tabs getting stuck on a loading screen
• Fixed full page reloads when navigating between pages in deployed apps

​

API & integrations

• Added Salesforce OAuth 2.0 Token Exchange support for authentication via Login Identity Provider
• Added BigQuery Workforce Federation authentication
• Fixed API errors silently failing — Clark now properly throws errors when an API run fails
• Fixed Databricks SQL queries getting corrupted when manually editing code
• Improved handling of integrations with large OpenAPI specs — metadata search no longer hangs
• Fixed Python code with escape sequences getting corrupted

​

Highlights

• Clark can now explore data when planning without asking for permission. Instead of prompting for permission to read, Clark can explore your integrations automatically at your direction to figure out the right approach. Query results appear right in the chat so you can see exactly what Clark is working with, drill in, and guide it toward the right queries.
• Clark tests APIs faster. We cut API test latency significantly — previously this could take 30+ seconds. Things should feel noticeably snappier when Clark is building and validating APIs for you.
• New safety protections. Clark now detects and blocks malicious or unsafe prompts, and we’ve added protections to prevent system prompt extraction attempts. These safeguards run automatically in the background.
• New integration: Databricks Lakebase. You can now connect to Databricks Lakebase directly from Superblocks.

​

Clark improvements

• Clark can now offer up to 6 options for multiple choice questions
• Fetched links are now clickable in web exploration
• Improved retry experience when Clark falls back across models and inference providers
• Clark now knows it cannot navigate between pages to take screenshots, reducing unnecessary loops

​

Deployed apps

• Opening app pages in new tabs now uses the correct URLs
• Cleaner deployed app URLs (removed deployed_at timestamp)

​

Editor experience

• Fixed issue where typing “ms” quickly would accidentally trigger preview mode
• Keyboard shortcuts like Cmd+Click now work correctly in fullscreen mode

​

Highlights

• Clark now stays smart over long sessions. Previously, Clark would lose context as conversations grew. Now Clark maintains a working memory—automatically retaining what was accomplished, key decisions, and constraints discovered.
• API generation is significantly faster and more accurate. We rewrote core parts of how Clark understands and generates APIs—full support for control flow features, a completely rebuilt parameterized SQL system, and improved handling of API documentation. Fewer mistakes, faster results.
• Inference errors no longer block you. Intermittent Vertex/Bedrock errors are now handled gracefully—if you see one, just send another message and Clark will recover.
• Clark can search the web. Clark can now perform real-time web searches. This is especially useful for using third-party libraries and looking up documentation for integrations in real time.

​

New capabilities

• Embedded apps with two-way communication. Superblocks apps can now communicate with parent apps and vice versa. Ask Clark to set up the interface for properties and events.
• Windows local development. Powershell support—no WSL required.

​

Clark generation improvements

• Clark can now build file upload features (e.g., send a file to an AI model or upload to bucket storage)
• Clark can now build apps with data tag switching
• Clark can now modify REST API body content types reliably
• Clark is better at targeted UI edits thanks to improved context about UI component layout
• Clark tests APIs faster
• Clark no longer hits timeouts when testing APIs

​

API generation

• Added support for all control flow features
• Rewrote parameterized SQL system—Clark can now always write parameterized SQL accurately and safely queries when parameterization isn’t possible
• Improved handling of API documentation and integration OpenAPI specs
• Fixed API transformation so Clark always sees user modifications
• Fixed missing bodyType field in REST API type declarations that caused TypeScript errors
• Added missing integration class declarations for 8 data sources (S3, MongoDB, GraphQL, and others)
• Prevented Clark from reading redundant JavaScript files in API directories

​

Chat experience

• Improved performance and reduced flickering for chats with many messages
• Chat no longer appears stuck thinking—now distinguishes between model thinking and time between tool calls
• Keyboard shortcuts for multichoice responses
• Fixed newlines not rendering correctly in AI-generated summaries
• Removed duplicate reconnection messages

​

Plan mode

• Plan mode always presents a new plan when you give feedback
• Clark has better context about targeted UI components and can make targeted edits more accurately

​

Reliability

• Edit mode loads reliably even after bad code generations, allowing Clark to fix issues
• Fixed chat crashes when clicking Clark’s interactive buttons (Build it, multiple choice) too quickly
• Fixed chat getting stuck centered on screen after generation
• Deployed app links now work correctly when opened in new tabs
• Syntax error indicator in top bar now clears after Clark fixes the issue

​

Versions & checkpoints

• Restore to any checkpoint from the Versions panel
• Restore to past checkpoint if you pause a generation midway
• Reset now works from the history panel even when unpinned

​

Other improvements

• Clone existing apps from the homepage
• Homepage search clears when entering folders
• Step navigation preserves previous run outputs
• Run Superblocks APIs locally without disabling browser local network access checks
• Changed “Think…” to “Starting development server” for clearer status during startup