Changelog
New features, improvements, and fixes in Skyvern
New features
- Settings Audit Log — Organization Admins now have an audit log in Settings with separate Changes and Runs tabs, 90-day filtering, CSV export, and cursor-based pagination, giving them one place to review configuration changes and run activity. (#9138)
Improvements
- Copilot Knows All Your Integrations — Copilot and the Integrations page now share the same list of available integrations, so asking Copilot which integrations you can use returns the same accounts — including vaults, MCP, and SMS — shown on the Integrations page.
- Specific Anti-Bot Error Codes — Runs blocked by bot protection now automatically report a specific error code (reCAPTCHA, hCaptcha, Cloudflare Turnstile, or a general anti-bot error) in the run’s errors list, without needing a custom error-code mapping.
Bug fixes
- Fixed an issue where another browser extension (such as a password manager) adding a frame to an agent’s tab could cause Skyvern to lose that tab mid-task instead of recovering it automatically.
- Fixed a regression that caused clicks to fail on some pages. (#9113)
- Fixed a race condition that could leave a run showing as still running in the runs list after it had already failed or been canceled. (#9122)
- Fixed file downloads being marked successful even when the response was an error page instead of the requested file. (#9114)
- Fixed download blocks reporting success when one of several requested files failed to download or save. (#9110)
- Fixed file download blocks taking up to 45 minutes to fail when the server returned a server error; they now fail fast.
- Fixed automations built with Copilot sometimes getting stuck on tabs and similar elements.
New features
- Drag-and-Drop Workflow Actions — Copilot can now build workflows that perform drag-and-drop actions between elements in the browser.
- MCP Feedback Tool — Added a feedback tool that lets agents using the Skyvern MCP report missing capabilities or unclear guidance directly to the Skyvern team.
- Copilot Access to Saved Bitwarden Logins — Copilot can now find and use logins saved in your connected Bitwarden vault instead of asking you to re-enter credentials you already have stored. (#9031)
Improvements
- Cleaner 1Password Import Categories — The 1Password import dialog now only shows item categories that can actually be imported as credentials, instead of letting you pick items that always fail to import.
- In-Chat Google Account Connections — Copilot can now connect Gmail (read-only) and Google Sheets accounts directly from the chat, without having to leave the page to visit Integrations.
- Runs List Date Filtering — The runs list API now supports filtering by a creation date range and defaults free-text search to the last 30 days to avoid timeouts for large organizations.
- More Reliable Secondary AI Tasks — Improved success rates and reduced costs for tasks like OTP parsing, image-to-shape conversion, and automatic title generation by upgrading the underlying secondary AI model.
- Cleaner My Agents List — Job application runs no longer clutter your My Agents list with duplicate auto-generated workflow entries.
- Copilot Error Context — Copilot can now use the recorded error from a page call that a completed block caught, helping it diagnose and repair workflows. (#8970)
- More Reliable Model Fallback — Improved task reliability by automatically retrying with a different model when a provider blocks or refuses a response, instead of letting the run fail. (#8971)
- Faster Dropdown Selection — Selecting an option from a long dropdown menu is now dramatically faster.
- Clearer Cloudflare Handling in New Sessions — Changed how the Cloudflare challenge auto-solver behaves in new browser sessions to avoid conflicting with Skyvern’s own Cloudflare handling.
- Clearer Proxy Failure Guidance — Failure messages for interrupted downloads over a residential proxy now suggest trying an ISP proxy instead.
- Faster Terminated Tasks — Tasks that end as terminated now finish noticeably faster, since a background check no longer delays the result.
Bug fixes
- Fixed the workflow editor’s Save button failing with an error when clicked while an authoring or Copilot action was still in progress; it’s now disabled until it’s safe. (#8961)
- Fixed the credential list showing a non-functional Edit option for read-only credentials imported from 1Password.
- Fixed a bug where automated logins on certain web apps built with Flutter could submit an empty password field.
- Fixed an issue where importing multiple 1Password items could leave some items partially imported if a later item in the same batch was denied.
- Fixed a 1Password credential refresh failing entirely when one vault was inaccessible, instead of continuing to update credentials from accessible vaults.
- Fixed reliability issues with automated form filling where a successfully submitted value could be incorrectly reported as failed on certain pages.
- Fixed an issue where refreshing or re-importing a 1Password credential could overwrite its previously verified website URL.
- Fixed download steps in AI-driven tasks so they finish correctly based on the files actually downloaded, instead of sometimes refusing to finish or crediting the wrong file.
- Fixed Copilot generating broken download steps for files located inside iframes, which previously caused the download to time out.
- Fixed an issue where unusually large or malformed PDFs could crash a workflow run; these files now fail quickly with a clear error instead.
- Fixed scheduled workflow runs failing to start when retrieving credentials from Bitwarden under heavy load.
- Fixed Copilot not being able to read the contents of SOP documents uploaded through the Upload SOP button or attached in chat.
- Fixed captcha solving failing when the system mistakenly tried to read a small icon instead of the actual captcha image.
- Fixed a bug where Skyvern could submit a form twice on slow-loading pages.
- Fixed an issue where Copilot could get stuck and refuse to show a new sign-in prompt after the user had already signed in once earlier in the same session.
- Fixed an issue where navigation blocks could save a blank model override, causing a misleading model indicator and inaccurate version comparisons in the workflow editor. (#8979)
- Fixed an issue where anti-bot protection could block automated browser workflows on certain websites.
- Fixed an issue where a self-hosted run could incorrectly start or save partial results after another server had already taken over its lease, which could leave runs in an inconsistent state. (#8982)
- Fixed an issue where clicking an answer tile in a form could be incorrectly reported as failed or get toggled back off. (#8983)
- Fixed an issue where a valid phone number was flagged as incorrect when confirming the entered value, if the field displayed the digits with special spacing.
- Fixed an issue where buttons inside certain embedded components, such as a download menu, were not detected by the agent, causing a task to finish without completing the intended action.
- Fixed an issue where filling in an autocomplete field could take several minutes and could be reported as successful even when no value was actually selected.
- Fixed an issue where an unresponsive page could cause an action to hang for several minutes instead of timing out. (#8990)
- Fixed an issue where a task could be forced to finish as failed upon reaching a verification code step, even when that wasn’t the task’s actual goal.
- Fixed an issue where processing certain PDF files could crash a worker and disrupt other running tasks.
- Fixed an issue where Skyvern could click the wrong element next to a small icon-only button instead of the button itself.
- Fixed the browser failure message shown when a run’s browser stopped responding not accurately describing what retrying will and won’t fix.
- Fixed an issue where a workflow block’s configured download timeout was ignored, causing large file downloads to be cancelled too early.
- Fixed an issue where Skyvern could treat an Arkose/FunCaptcha challenge as solved before it was actually cleared, causing login flows to stall.
- Fixed workflow runs failing when retrieving a Bitwarden API-key credential took longer than expected, by giving the unlock step more time to complete.
- Fixed an issue where Skyvern could give up waiting for a verification code too early because a normal “no code yet” response was misread as an error. (#8997)
- Fixed the “Browser error” tooltip promising a fresh browser on retry even when a run was reusing an existing browser session or connecting to your own browser.
- Fixed workflow runs triggered by external events being permanently rejected after a temporary submission failure instead of being automatically retried. (#8999)
- Fixed an issue where a run’s action history could show a download as completed even though it actually failed.
- Fixed Skyvern giving up right as a site’s acceptance response was about to arrive, causing false failures on job application workflows.
- Fixed an issue where runs that successfully saved their file could still show an unrelated download error.
- Fixed an issue where an unresolved CAPTCHA challenge on a page could block Skyvern from handling other CAPTCHA types also present on that page.
- Fixed job application workflows not using their one permitted retry after a site rejected a submission, causing false failures. (#9003)
- Fixed an issue where canceling passkey setup before it finished could leave behind a credential that could never be deleted.
- Fixed a bug where Copilot could get stuck on “checking for saved changes” and lock the chat after a change failed to save. (#9006)
- Fixed an issue where an interrupted file download during an automated task could be saved and treated as a completed file instead of being discarded. (#9009)
- Fixed an error that could prevent a completed data-extraction task from loading correctly or delivering its completion webhook. (#9016)
- Fixed an issue where solving certain captchas could hang a task run for hours instead of timing out properly.
- Fixed an issue causing some automated task steps to fail when the AI model used its entire response budget on internal reasoning.
- Fixed an issue where a failed page navigation in a persistent browser session could surface as a generic “browser unavailable” error instead of the real underlying error. (#9020)
- Fixed workflow runs continuing on after an unexpected browser disconnection and risking an unverified success instead of failing immediately. (#9022)
- Fixed an issue on pages with many frames where duplicate internal frame identifiers could cause automation to interact with the wrong element.
- Fixed an issue where Skyvern kept re-entering the same externally supplied one-time passcode after a site rejected it instead of waiting for a new one. (#9026)
- Fixed an issue where a legitimately downloaded CSV file could be mistakenly rejected as an error page during a task run. (#9029)
- Fixed an issue where some task workflow runs were missing their final screenshot in the run details. (#9030)
- Fixed a race condition in anti-bot challenge handling that could cause a browser automation task to hang instead of properly timing out.
- Fixed MCP tool calls with timeout values outside the allowed range being silently rejected or misread instead of returning a clear error message.
- Fixed an issue where the analytics dashboard could incorrectly show zero activity for days that actually had completed runs.
- Fixed an issue that caused every tool call from newer MCP clients, including Claude, to fail after the initial connection handshake.
- Fixed tasks giving up early instead of retrying as configured when a failure was mistakenly flagged as anti-bot detection.
- Fixed task runs repeatedly retrying against a dead browser instead of stopping and reporting failure immediately when the browser was lost.
- Fixed an issue where Yes/No style question buttons on application forms could record the answer to the wrong question. (#9037)
New features
- Email Attachment Downloads — Workflows can now optionally download email attachments found by the Email Inbox block, so later steps can use those files directly.
- Per-Block Analytics Breakdown — The agent analytics page now shows a breakdown of which workflow block an agent’s actions and failures come from, making it easier to pinpoint where an agent is slow or flaky. (#8902)
- Event Triggers in Studio — Added a Triggers panel in Studio for creating, managing, rotating, and monitoring signed event trigger integrations, including delivery history and replay. (#8933)
- 1Password Vault Import — Added the ability to import selected items from an organization’s 1Password vault into Skyvern credentials, with a dedicated import UI and support for keeping them refreshed. (#8952, #8953)
Improvements
- AWS WAF Captcha Solving — Skyvern can now automatically solve AWS WAF “confirm you are human” captcha challenges, not just Cloudflare ones.
- Icon-Only Button Recognition — Skyvern can now recognize and click icon-only buttons, such as edit or delete icons, that have no visible text label.
- Iframe Captcha Handling — Improved captcha handling on privacy-request forms that embed a verification widget in an iframe, so legitimate submissions are no longer incorrectly flagged as bot traffic.
- Self-Hosted Intranet Access — Self-hosted deployments can now disable URL validation for internal or intranet targets when their network policy requires it. (#8881)
- Copilot Network Failure Detection — Copilot can now detect and report when a page’s own network request fails, instead of appearing to wait on it indefinitely.
- More Accurate Yes/No Question Detection — Improved accuracy of Yes/No question detection on job application forms, reducing the chance of submitting an incorrect answer to questions like work authorization or sponsorship.
- More Reliable Job Application Sign-In — Job application recipes now handle sign-in and emailed verification codes more reliably, improving how often automated applications complete successfully. (#8897)
- Sharper Copilot Screenshots — Copilot now sends screenshots at a sharper resolution with each message, improving its ability to read small text and fine details on a page. (#8905)
- Clearer Copilot Sign-In Label — Copilot’s credential card now clearly labels “Sign in myself” as a one-time option for the current test, rather than implying it saves your sign-in for future runs.
- Cursor-Based Task List Pagination — Added optional cursor-based pagination to the task list API for faster, more reliable paging through large task histories.
- Copilot Page-State Refresh — Copilot now refreshes its view of the page after every browser automation step, so it builds and fixes workflows based on the page’s current state instead of an outdated screenshot.
- Visible Run Error Codes — The run Overview page now shows a failed run’s error codes directly, without needing to open the Outputs tab.
- Up-to-Date Model for Older Workflows — Workflows created before 2026 now run on the same up-to-date AI model as newer workflows instead of being held back on an older one.
- Faster Code Block Import Checks — Workflow code blocks now catch disallowed imports immediately when you save them, and support common built-in imports, instead of failing partway through a run.
- Structured Expired-Session Errors — Responses for expired browser sessions now include a structured error code, and run summaries show a clearer path to starting a new session.
- Faster Permanent Delivery Failure Reporting — Event trigger deliveries that fail for a permanent reason, such as missing configuration, are now reported immediately instead of being retried repeatedly before failing.
- Faster Task and Run Listing — Listing tasks and runs through the API now responds noticeably faster.
- Concurrent Webhook Delivery — Webhook and event-trigger deliveries are now dispatched concurrently, reducing delivery delays during traffic bursts.
- Clearer 1Password Empty-Vault Guidance — Added clearer in-app guidance on the Passwords tab when a connected 1Password account has no listable login items.
- Retryable Browser Session Limits — Browser session creation now returns a quick, clearly retryable response instead of timing out when an organization starts too many sessions at once.
- Longer Copilot Conversation Memory — Copilot can now retain more conversation history before trimming, reducing cases where it lost track of recent actions during long chats. (#8958)
Bug fixes
- Fixed an issue where a scrolling screenshot could be dropped entirely if the page scrolled backward near the end of the capture.
- Fixed an issue where certain job-application workflow runs would fail to complete even when they did not actually require a saved credential.
- Fixed an issue where Copilot would refuse every further action in a conversation after a single conflicting edit, instead of recovering automatically.
- Fixed an issue where Copilot’s scheduling and account tools could be blocked by a pending workflow draft the user could no longer see.
- Fixed an issue where canceling a run could still let an in-progress dropdown selection complete and change the page afterward.
- Fixed file parsing (PDF and image extraction) hanging indefinitely instead of failing with a clear error when an extraction call stalled.
- Fixed large-document text extraction failing prematurely by increasing the allowed processing time. (#8888)
- Fixed Copilot sometimes ending a turn with no response at all when it reached its time limit.
- Fixed job application runs sometimes being marked as a plain failure instead of correctly recognizing that a posting had been removed or closed.
- Fixed scheduled runs, code blocks, cached steps, and SDK actions running first and being billed afterward instead of being blocked when an account was already out of credits. (#8894)
- Fixed Copilot marking test runs as stopped when a turn’s time limit was reached, instead of clearly showing in chat that they were left running.
- Fixed phone number fields that misread a national number as an international one, causing Skyvern to repeatedly retype the value instead of asking once, clearly, for the country code. (#8896)
- Fixed login flows occasionally losing an already-fetched emailed or texted verification code and getting stuck instead of reliably using it. (#8898)
- Fixed document extraction returning incorrect or inconsistent values when a PDF’s text contained corrupted characters; it now falls back to image-based reading.
- Fixed job application forms incorrectly flagging correct radio-button and checkbox answers, such as work authorization questions, as needing to be changed. (#8899)
- Fixed a race condition when starting a new Copilot chat while chat history was still loading; the “New chat” button is now disabled until loading finishes. (#8900)
- Fixed workflow runs waiting on a human interaction step either waiting out the full timeout or being marked successful incorrectly when the browser session was closed; they now end immediately with the correct status.
- Fixed date-picker fields that take many steps to resolve getting stuck for 20-30 minutes; Skyvern now works through them within a fixed time budget.
- Fixed searches of a workflow’s run history occasionally timing out with a generic server error instead of returning quickly with a clear error.
- Fixed an issue where Copilot could fail to confirm that a change you accepted was actually saved. (#8906)
- Fixed analytics breakdowns by workflow tag or metadata undercounting runs for the current day before data had fully landed.
- Fixed a bug that could cause a run to use the wrong AI model when multiple runs executed on the same worker at once. (#8908)
- Fixed file-download workflow blocks failing to retrieve files when running through certain anti-bot-protected browser sessions.
- Fixed large artifact downloads failing or timing out by streaming file content instead of loading it entirely into memory.
- Fixed workflows getting stuck on TrustArc consent forms by automatically retrying the submission after a bot-check rejection.
- Fixed Copilot sometimes reading a hidden duplicate of a value instead of the visible one when building a workflow.
- Fixed new Google Cloud Marketplace purchases failing to get approved.
- Fixed Copilot-built workflows breaking when a page’s visual styling changed, by having Copilot identify repeated labels by their content instead of their styling.
- Fixed Copilot occasionally proposing an untested workflow change when it ran out of time while building a response.
- Fixed failed workflow runs showing a generic recovery error instead of the real underlying error when automatic recovery also failed.
- Fixed performance slowdowns in browser automation caused by frequent pauses after multi-viewport screenshots.
- Fixed Copilot-generated code failing with an execution-context error when reading a value immediately after a page navigation.
- Fixed an issue where sensitive values from a run could appear in the self-heal recovery details when a code block’s automatic fallback failed.
- Fixed an issue where tasks targeting a city with no available proxy would fail outright instead of falling back to a proxy in the same country.
- Fixed an issue where a Copilot conversation could hang indefinitely and never complete if the underlying model connection stalled.
- Fixed a stopped Copilot task incorrectly reporting that it ran out of time instead of showing it was stopped at your request.
- Fixed searches of your organization’s runs hanging and failing after a long delay instead of returning a clear error quickly.
- Fixed self-hosted deployments leaving runs stuck running forever after the server crashed or restarted, instead of failing and notifying callers. (#8925)
- Fixed an issue where Copilot chat messages could occasionally appear out of order in the conversation history.
- Fixed Flutter web apps appearing blank or failing with a misleading page-load error; Skyvern can now see and interact with them normally. (#8928)
- Fixed an error that could cause workflow submissions using a shared browser session to fail unexpectedly when submission took longer than usual.
- Fixed workflows that explicitly disable AI fallback still triggering an AI-powered recovery attempt when a code block failed.
- Fixed plan tier and credit figures in Analytics failing to load for non-admin organization members.
- Fixed an issue where runs started from the web app could occasionally be routed to a slower, lower-priority model tier, causing noticeable delays.
- Fixed a bug in the workflow builder where opening and closing a loop or conditional block’s editor panel incorrectly triggered an “Unsaved Changes” prompt. (#8934)
- Fixed an issue where filling in a form field could time out if the field briefly became read-only while automation was interacting with it.
- Fixed a timed-out Search workflow block request immediately failing the entire run instead of being automatically retried.
- Fixed an issue where checking on a very large workflow run could return no information at all, including losing the links needed to download completed files.
- Fixed job application runs mislabeling being blocked by a site’s bot protection as a genuine login failure, and vice versa. (#8946)
- Fixed repeated failures caused by a single bad browser connection; Skyvern now automatically takes a connection out of rotation after it repeatedly fails to load pages.
- Fixed the Studio Triggers panel showing incorrect local times for deliveries and occasionally discarding unsaved trigger mapping edits on refresh.
- Fixed an issue where saving an invalid 1Password token could silently break an already-working 1Password connection.
- Fixed a bug where pages with looping iframes could hang a browser automation run; affected runs now fail cleanly instead of stalling indefinitely.
- Fixed one-time verification codes failing to submit when a login page splits the code across multiple input boxes.
- Fixed autocomplete on dropdown-based form fields stalling or submitting an incomplete value when no options were initially available.
- Fixed an issue where a canceled workflow run could send its final webhook notification twice.
- Fixed an issue where deleting a workflow or folder could leave its scheduled runs active, causing them to fail repeatedly.
- Fixed imported 1Password credentials to show clearer, specific error messages when generating a one-time verification code fails. (#8951)
- Fixed an issue where Copilot could run two conflicting actions in the same chat at once after a dropped connection, page reload, or duplicate message. (#8955)
- Fixed an issue where Copilot could repeatedly request a new screenshot without ever actually seeing one during long conversations. (#8956)
Improvements
- Faster Large-Output Run Pages — Run detail and timeline pages now load faster for runs with very large output values.
- Subscription Cancellation Email — Customers now receive an email confirmation when they cancel their subscription, including their access end date.
- Copilot Selector Match Visibility — Copilot now shows how many elements a selector matches and whether a click will actually reach a form control, helping it build more reliable automations.
Bug fixes
- Fixed validation workflow steps reloading or navigating away from a page they were already on, causing false failures on session-ticket pages.
- Fixed legal and compliance answers typed with Enter being submitted before they were checked for support, which could let an unsupported answer go through.
- Fixed radio-button and checkbox answers to legal questions, like work authorization, not being checked for support before a form submitted, unlike typed answers.
- Fixed workflow steps that successfully changed the page, such as toggles or filters, being incorrectly marked as failed.
- Fixed contract-billed organizations getting stuck after a subscription cancellation and losing access to the Manage Subscription option; billing now repairs automatically.
- Fixed the workflows list timing out while loading reliability badges for organizations with large numbers of historical runs.
- Fixed a Copilot workflow build that recently failed testing still showing as Tested on later review turns.
- Fixed the Tested indicator on a Copilot-reviewed workflow block showing an outdated test result instead of the most recent one.
- Fixed the runs list timing out when filtering runs by error code.
- Fixed the job-application session list reading “Submitted this session,” which implied an application was received even when it had failed or was still starting; it now reads “Started this session.”
- Fixed free-text searches of the runs list timing out instead of failing fast with a clear error.
- Fixed Copilot silently signing in with a different saved login when the one you named couldn’t be found; it now asks for guidance instead.
- Fixed runs sometimes incorrectly flagging a valid credential as rejected; invalid-credential results are now verified before being reported.
- Fixed task agents judging a task complete before a page’s content finished rendering, causing premature completions on pages that render content after load.
- Fixed one-time login codes typed during a run being visible in action history and run details.
- Fixed an issue where SDK-triggered browser actions could lose their session partway through if the action took a while to complete.
New features
- Downloadable Copilot-Generated Files — Secure CodeBlocks and Copilot can now publish generated XLSX, PDF, CSV, and ZIP files as run downloads and chat download buttons. (#8693)
- Clipboard Paste in Live Browser Control — You can now paste your clipboard while in control of a live browser, from one control bar shared by CDP and VNC sessions. (#8699)
- Redesigned Discover Home — The Discover page’s prompt, examples, and templates have been redesigned.
- Guided Full-Page Welcome for New Sign-Ups — New accounts now land on a full-page walkthrough covering goals, expected volume, and role, replacing the old setup modal.
- Guided Template Walkthrough — Using a home page template now copies it, opens Copilot with a numbered summary of what it does, and shows an inputs card for the credentials and values it needs. (#8730)
- Copilot-Extendable Browser Sessions — Ask Copilot to extend its own signed-in browser session instead of only being able to do it through the API, up to the existing 6-hour cap.
- Terminate Block Error Codes — The Terminate block can now carry a custom error code, so retry rules and webhook consumers can match on it instead of parsing the free-text reason.
- Google Cloud Marketplace Purchases — Orders purchased through Google Cloud Marketplace now provision, change plan, suspend, reactivate, and cancel automatically.
- First-Run Product Tour — New users now get a guided walkthrough of the Copilot and Browser panes on their first run, replayable anytime from Settings.
- Copilot-Saved Browser Profiles — Ask Copilot to save a closed signed-in browser session, or a run, as a reusable named browser profile.
- Activation Credits — New users now earn activation credits for onboarding steps, shown in a credits meter on the welcome page and Home checklist.
Improvements
- Google Sheets Range Support — Copilot now resolves a Sheets tab from a URL’s
gidat save time and shows the Range in append mode. (#8696) - Clearer Personal Account Naming — A personal account is now named after its Skyvern organization in the org menu and settings sidebar.
- Recording Controls — The Copilot recording sticky bar is now labeled and includes a stop control.
- Redesigned Prompt Box Settings — Advanced prompt box settings now live in a gear popover.
- Clearer Credential Refresh — The saved-login refresh link now explains what it does.
- Clearer Copilot Change Review — Copilot now shows its proposed changes on one annotated canvas, with plans and credential receipts shown where they happened in the turn.
- File Input URLs for Copilot — Copilot now knows a file input can accept a user-supplied URL, not just an upload.
- Pinned Copilot Questions — Pending Copilot questions now dock above the composer instead of scrolling out of view.
- Flexible Copilot Answers — Copilot questions can now be answered in your own words instead of only the offered choices. (#8713)
- Chronological Copilot Activity Log — Copilot’s activity log now reads in the order things happened, with redesigned answered-question cards. (#8718)
- More Resilient Workflow Studio — Workflow Studio now contains pane crashes instead of taking down the editor.
- Skyvern 3.0 Recommended — The Skyvern 3.0 engine badge is now labeled “Recommended” instead of “Beta”.
- Code Block Goal Editing — Editing a code block’s Goal now keeps your text and offers to update the block to match. (#8711)
- Faster Web Search — Copilot and code blocks now search the web through a dedicated search API instead of a browser tab. (#8708)
- Clearer Billing Errors — Skyvern now shows the reason when a billing plan change fails.
- Faster Credential Failures — A login block now fails fast with a clear reason when it isn’t bound to any of the workflow’s credentials, instead of opening a browser first.
- Copilot Test Results as Cards — Each block Copilot tests now shows as one card with its status, code change, and failing line.
- Copilot Run History by Status — Copilot can now list a workflow’s recent runs filtered by status, such as failed or terminated, instead of only seeing the most recent run.
- Longer Task Runs on Long Forms — Task runs now get more time to finish long forms that are still making visible progress, instead of stopping once the token budget runs out.
- Unsubscribe from Skyvern Emails — You can now unsubscribe from categories of email Skyvern sends, like the analytics digest, with a one-click link.
- Default Engine Option in Workflow Editor — The block engine picker now has a “Default” option separate from Skyvern 1.0 and 2.0, which are now labeled Legacy. (#8735)
- Docked Copilot Sign-In Requests — Copilot’s sign-in and Google account requests now dock above the chat input instead of scrolling out of view with the transcript.
- More Reliable Copilot Repairs on SSL Errors — Copilot no longer gives up early on ambiguous SSL connection errors during automatic repairs; it now attempts a fix. (#8731)
- Upstream Output Reuse in Copilot Repair Tests — Testing a repaired block that reads an earlier block’s output no longer fails with a template error; the test now reuses that output automatically. (#8737)
- Faster Copilot Repair Tests — Copilot now reuses outputs from already-completed steps when testing a fix, instead of rerunning them.
- Settings Header Alignment — The Settings page header now aligns with the navigation rail instead of sitting in its own row above it.
- Clearer Run Start Time — The Timeline now labels a run’s start time as “Created” instead of a timestamp that could be mistaken for when the run finished.
- More Reliable Date Field Filling — Task runs now tab out of certain hard-to-click date fields before reading their value back, fixing cases where a filled field was incorrectly reported as empty.
- Guided Credential Example on Home — The “Log into an HR portal” example on the Home page now walks new users step by step through connecting and using a demo login.
- Copilot Names Agents from the Home Prompt Box — Agents created from the Home page prompt box now get a short, Copilot-generated name instead of keeping the full prompt as their title.
- Smarter Partial Date-of-Birth Handling — When a task only provides a birth year, Skyvern now enters
01/01/<year>instead of terminating, and never invents an address or phone number. (#8747) - More Reliable Dropdown and Combobox Selection — Improved how Skyvern verifies that a dropdown list was read completely and that a selection was actually committed, reducing false “not found” and “didn’t select” errors.
- Behavior-Based Onboarding Emails — Onboarding emails now adapt to what a new user has actually done, and free accounts are notified when they run out of credits.
- Visible Conditional Block Errors — The run view now shows when a conditional block’s condition couldn’t be evaluated and which branch was taken, instead of showing nothing.
- Enter to Send on the Home Prompt Box — Press Enter to send a prompt and Shift+Enter for a new line, matching Workflow Copilot’s chat input. (#8741)
- File Upload on the Classic Home Prompt Box — Added the ”+” upload menu to the original Home page prompt layout, so you can attach files there too.
- More Accurate Code Block AI Fallback — The AI fallback for code blocks now sees the exact values the failed code was typing, instead of guessing from placeholder text.
- Goal Update Prompts on Manual Code Edits — Hand-editing a code block’s code now offers to update its Goal to match, or keep your own wording.
- Skyvern 3.0 Listed First — The block engine dropdown now lists Skyvern 3.0 (Recommended) above the legacy 1.0 engine.
- Low-Credit Warning Email — Free accounts now get a one-time email when their free credits run low, instead of finding out only after they run out.
- Simplified Onboarding Emails — Added a Day-1 email introducing Copilot, simplified the Welcome email to two steps, and stopped sending “no successful run yet” nudges to team accounts.
- Welcome Questionnaire for New Sign-Ups — New sign-ups now land directly on the welcome questionnaire instead of briefly landing on Home first.
- Clearer Insurance Quote Example — The Home page’s insurance quote example now reads as plain text instead of a JSON blob.
- Clearer Extraction Results — Skyvern 3.0 extraction-only blocks now always report exactly what’s on the page, marking any missing field as null. (#8771)
- Instant Welcome Email — New signups now get the Welcome email right after their first app load, instead of waiting up to an hour.
- Clearer Copilot Loop Debugging — Workflow Copilot now shows which input a for_loop block actually used, with recorded row counts, making a failed loop easier to debug.
- Readable Copilot Browser-Code Steps — Copilot’s browser-code tool calls now list the steps the code performed, like “Opened a page → clicked next → read ×10,” instead of just “Ran browser code.”
- Clearer Blocked Code-Block Errors — Copilot now explains which value caused a blocked code block and how earlier outputs reach code, instead of a generic error.
- Visible Page-Load Failures in Code Blocks — A CodeBlock wait-timeout error now shows the page load failure that caused it, instead of just the timeout.
- More Complete Copilot Repairs — Copilot’s automatic repairs now see icon-only controls, like expanders and chevrons, even when their selector isn’t unique on the page.
- Starter-Prompt Onboarding Email — New signups now get an email with three starter prompts — collecting website data, downloading invoices, and filling out a form — to help them get going.
- Readable Copilot Tables — Copilot replies now render Markdown tables, headings, quotes, and dividers properly instead of as unstyled text.
- Search Runs by More Identifiers — Run History and run search can now find runs by browser profile ID, browser session ID, credential ID, or webhook callback URL, not just run ID or inputs.
Bug fixes
- Fixed the extension bridge processing page events out of order relative to earlier responses. (#8694)
- Fixed bot-challenge pages being reported as “no challenge” instead of correctly detected. (#8695)
- Fixed credential approvals being lost for credentials used past the recording evidence size cap.
- Fixed run errors losing their mapped error code when the task also had an OTP timeout.
- Fixed Copilot forgetting an accepted state across page reloads.
- Fixed a closed Workflow Studio pane reopening at the wrong edge instead of where it was.
- Fixed Copilot comparing proposals against a stale staged draft instead of the current canvas.
- Fixed edits being lost when rejecting a Copilot proposal that only renamed the workflow or never reached the canvas.
- Fixed Gemini rejecting the MCP tool list due to a missing schema on an output parameter.
- Fixed the extension relay stopping when an event callback raised an error.
- Fixed run failure messages leading with an outer error instead of the actual cause.
- Fixed false “Copilot is locked” toasts when expanding chat steps or switching editor views.
- Fixed runs failing at max retries instead of terminating with the correct mapped error code.
- Fixed plan upgrades not rolling back correctly if an in-progress ACH payment failed.
- Fixed the self-hosted Discover page crashing on sign-in hooks.
- Fixed literal “N/A” and “NULL” text being dropped when parsing Excel files.
- Fixed PDF generation failing permanently after a page-renderer crash; it now retries on a replacement tab.
- Fixed clicks on natively disabled controls taking up to 15 seconds to fail; they now fail in 2 seconds.
- Fixed the AI fallback for code blocks recovering at the wrong address for a wrapped or indented
goto. - Fixed concurrent Copilot test runs overwriting each other’s workflow version.
- Fixed read-only and disabled fields reporting a successful fill or clear instead of being refused.
- Fixed agents with no blocks being runnable.
- Fixed the Recording and Screenshots pills showing on finished runs that have neither.
- Fixed the onboarding questionnaire not reappearing after a page refresh. (#8706)
- Fixed a recording refinement draft being presented as a validated workflow before it was. (#8712)
- Fixed Skyvern 3.0 refusing a re-ask conversion whenever a settled page’s layout shifted at all, instead of only when the underlying document changed.
- Fixed Copilot’s Review view landing on a pending-proposal card with no available actions. (#8714)
- Fixed the Copilot review-gate buttons wrapping in a narrow pane and its Test end-to-end action not being locked like Reject. (#8715)
- Fixed Skyvern 3.0 mishandling a covering layer over the target field and mis-scoring an action block whose one action succeeded but whose follow-up was refused by the step cap. (#8722)
- Fixed one-shot downloads not recovering correctly after a server error.
- Fixed the organizations page showing a blank state instead of a full-page retry option when it failed to load.
- Fixed Workflow Studio showing replay pills before there was anything to replay.
- Fixed a Cloudflare Turnstile checkbox being reported as solved after it silently reset, instead of being re-clicked.
- Fixed browser pages hanging for other users sharing a browser session after one person disconnected mid-interaction. (#8736)
- Fixed Copilot failing before sending any request when using certain GPT-6 models.
- Fixed task runs refusing to click a checkbox that sits on top of its own styled replacement, reporting it as not rendered. (#8734)
- Fixed task runs losing track of a page after a click opened a new tab, causing the run to stall. (#8733)
- Fixed a dark-mode flash of a white screen on sign-in, refresh, and other full page loads.
- Fixed a shared browser session hanging for other users when closing one client’s connection failed to cleanly detach.
- Fixed the Settings organization profile panel showing excess empty space below its content.
- Fixed Copilot’s action explanations sometimes appearing next to the wrong action.
- Fixed a run’s error details sometimes appearing twice due to a duplicate termination error record.
- Fixed a toolbar flicker on first load of the Home page prompt box.
- Fixed Copilot being unable to read large pages after signing in with a saved login, and it now tells you when a page is too large to show. (#8742)
- Fixed data extraction failing partway through on pages with many rows due to hitting a browser handle limit.
- Fixed cached workflow runs executing every branch of a conditional block inside a loop instead of only the correct one.
- Fixed Copilot failing with a generic connection error after an Anthropic Bedrock request was denied, instead of falling back correctly.
- Fixed automatic error recovery sometimes following outdated step instructions instead of the block’s actual goal, which could cause it to submit incorrect information.
- Fixed Copilot showing a misleading “worker unavailable” message when a test run failed to prepare.
- Fixed the code block AI fallback running during Copilot build tests and editor block runs, hiding real code regressions; run details now show which block it rescued and why the code failed.
- Fixed a Terminate block’s custom error code or reason being able to leak a referenced secret’s value.
- Fixed task runs occasionally being marked failed or terminated right after a successful form submission.
- Fixed Skyvern 3.0 refusing to finish a task when the page had a live countdown, clock, or rotating headline.
- Fixed Skyvern 3.0 losing its view of the page entirely on sites that define their own
toJSON, instead of reading the page correctly. (#8770) - Fixed Skyvern 3.0 occasionally picking the wrong row from a long dropdown or menu due to a shifted or guessed row number. (#8767)
- Fixed a single-action step sometimes being marked complete before its action actually took effect, which could carry the run onto the wrong next page.
- Fixed Skyvern 3.0 missing icon-only close buttons on overlays and incorrectly reporting that no controls were found. (#8766)
- Fixed Copilot’s automatic repair discarding the notice that a full-screen overlay, like a consent banner, was blocking the page.
- Fixed Skyvern 3.0 on slow-loading pages misjudging whether the page had actually finished loading.
- Fixed Skyvern 3.0 incorrectly reporting a search-and-select field as empty when a single result committed automatically.
- Fixed Skyvern 3.0 clicking the wrong control when its screen-reader label differed from its visible text.
- Fixed Copilot failing before sending any request on certain Anthropic and Bedrock models.
New features
- Terminate Block — A new workflow block lets a run end early with a custom reason, instead of running through remaining blocks.
- Copilot-Managed Schedules — Copilot can now create and manage the open workflow’s recurring schedules directly from chat.
- SMS Two-Factor Authentication — Credentials now support SMS-based two-factor authentication with managed phone numbers.
- Elapsed-Interval Schedules — Workflow schedules can now run on a fixed interval measured from an anchor time, instead of only fixed calendar times. (#8660)
- One-Time Scheduled Runs — Workflow schedules can now be set to run once at a specific date and time. (#8680)
- Copilot File Upload to Pages — Copilot can now upload a file you attach in chat directly into a page’s file picker. (#8691)
Improvements
- Full YAML Workflow Editing — The workflow editor’s YAML view now exposes and accepts every workflow setting.
- Declared Code Block Return Schema — Code blocks can now declare a return schema for Copilot, and AI-fallback output is recorded as the block’s output.
- Optional Extension User Scripts — Browser control no longer requires the User Scripts permission to be enabled.
- Copilot Repair Test Accuracy — Copilot repair test runs now reuse the original failed run’s inputs.
- Clearer Billing Navigation — The sidebar “Add credits” button is now labeled “Upgrade plan” / “Plan & usage”.
- Organization Terminology — Settings, the org menu, and analytics banners now say “Organization” instead of “Workspace”. (#8697)
- Agent Creator Visibility — The agents list now shows each agent’s original creator.
- Agent Creation Date — The agents list now shows each agent’s first-version creation date.
- Local File Reads in Blocks — Blocks can now read a local-storage upload referenced by a
file://URI. - Copilot Page-Challenge Recovery — Copilot can now solve a page challenge or start a fresh browser when scouting a page.
- Clearer Copilot Run Results — Copilot now reports how it picked a run for results and lists newer finished runs.
- Browser Profile Preview — The workflow editor now shows a read-only preview of the browser profile a run would load.
- CAPTCHA Support in Code Blocks and Copilot — Code blocks and Copilot browser-code cells can now solve image-text CAPTCHAs directly.
- More Resilient Browser Recordings — Record Browser no longer rejects a recording outright; it now emits repair steps for actions it can’t replay.
- Recording Visibility in Copilot Chat — Browser recordings now show as a live chapter inside the Copilot chat.
- Save Credential on Enter — Pressing Enter in the password field now saves the credential.
- Chat During Recording — You can now talk to Copilot while a browser recording is capturing.
- Credential and Session Ownership — Credentials and browser sessions now show who created them. (#8654)
- Cached Script Retention — Data-retention cleanup no longer removes cached scripts, keeping optimized runs fast.
- Customer Instructions Take Priority — Under a new run setting, Skyvern 3.0 now lets your workflow instructions override its general behavior rules, and no longer lets page content override customer-provided values.
- Layout-Resilient Cached Code — Saved code blocks can now target metric cards with a label-anchored selector, so they survive page layout changes.
- Cleaner Run Timelines — Run timelines now show an action’s result only when one was actually recorded.
- Plain-Language Failure Labels — Run failures now show plain-language labels instead of raw internal category names.
- More Reliable CAPTCHA Solving — Skyvern now remembers which CAPTCHA type a given site rejects, improving solve reliability on protected pages. (#8657)
- Clearer Browser Block Validation — The first browser block’s URL is now required and validated before the block runs.
- Queued Message Indicator — A queued Copilot message now shows as a strip on the message box. (#8659)
- Quieter Browser Pane — The browser pane header is now less cluttered, with an overflow menu, run-only replay tabs, and clearer browser status. (#8668)
- Fewer Extension Permissions — The Skyvern Agent browser extension no longer requests the User Scripts or unused host permissions.
- Search Block Improvements — The Search block now uses standard error messages, and a Data Schema can be used without a Prompt. (#8679)
- Copilot Test Accuracy — Copilot build tests now start from the saved browser profile.
- Copilot CAPTCHA Reading — Copilot can now read image-text CAPTCHAs while scouting a page and in saved code blocks. (#8683)
- Run Attribution — The Runs page now shows who started each workflow run. (#8677)
- Clearer Copilot Status — Copilot now shows when it’s working or waiting on you above the message box, and the Stop button matches Send.
Bug fixes
- Fixed the workflow editor losing your exact state after a rejected Copilot proposal; saves are now transactional.
- Fixed one-time-code entry: Skyvern now retries once with a fresh code when a site rejects a multi-box code, and types full verification codes into one-character boxes reliably. (#8717)
- Fixed the Copilot credential card staying open after picking a saved login on a new chat.
- Fixed intermittent loss of live-browser control by re-sending take-control on VNC reconnect and opening the live-view socket only after VNC connects.
- Fixed a sent Copilot prompt occasionally staying stuck in the queue.
- Fixed several Skyvern 3.0 typing issues: appended text is now reported and placed correctly, and an offered autocomplete row is committed when none exactly matches the typed value.
- Fixed clicks into embedded frames failing when the frame hadn’t finished rendering.
- Fixed the Copilot activity log losing a finished step’s title when new work started right after.
- Fixed Skyvern 3.0 hiding a read of a different browser tab as if it were a repeated read.
- Fixed Skyvern 3.0 accepting task completion on a blank page or without checking the result against the stated goal.
- Fixed Copilot sending a canned message instead of a real reply when a tool call produced an empty model response.
- Fixed credentials chosen during a browser recording not carrying into its refinement turn.
- Fixed Skyvern 3.0 hiding or mislabeling clickable rows, menu options, and buttons combined with clickable containers during page observation. (#8716)
- Fixed Skyvern 3.0 not generating a real code for a credential’s TOTP placeholder, unlike the previous engine.
- Fixed dropdown search fields that only filter on Enter, so Skyvern 3.0 can search and filter windowed dropdowns correctly.
- Fixed proxy tunnel and SOCKS failures being mislabeled as navigation failures.
- Patched a security vulnerability in TLS certificate hostname validation used by outbound requests.
- Fixed extension-bridge tabs and page reads breaking when a page changed its URL.
- Fixed code blocks failing as busy under load instead of waiting for a free runner slot.
- Fixed Copilot forgetting answers to its own questions after a few more tool calls.
- Fixed Skyvern 3.0 writing dates into the wrong segment of a segmented date field.
- Fixed Copilot’s turn not finishing recovery correctly when a stream loss occurred but the turn ended normally.
- Fixed the Search block returning results outside the requested site filter or missing results with no cached content.
- Fixed Copilot showing a false “unsaved changes” warning on retry.
- Fixed Reject retries in Copilot chat leaking across chats instead of staying scoped to the chat they came from.
- Fixed the browser extension grouping tabs together in popup windows.
- Fixed data extraction nulling out fields the goal asked to produce, instead of only nulling fields not present on the page.
- Fixed Skyvern 3.0 skipping later toggle clicks in a batch after an earlier one failed.
- Fixed code block page requests bypassing the run’s configured browser proxy.
- Fixed Skyvern 3.0 refusing to re-enter a login username the page had overwritten, while keeping the existing cap on password re-entry.
- Fixed action steps being charged for calls Skyvern 3.0 refused before acting on the page. (#8688)
- Fixed Copilot’s first workflow submissions failing on malformed syntax; they’re now built from structured objects.
- Fixed a terminated Skyvern 3.0 block not getting a retry when the site skipped a screen its criterion expected. (#8687)
- Fixed Copilot losing rows past the 50,000-row cut when paging a long run’s results. (#8684)
Improvements
- Workflow API settings — PUT updates keep stored values when you omit
extra_http_headers,cdp_connect_headers,totp_identifier,totp_verification_url,webhook_callback_url, orproxy_location. Explicitnullclears these settings. A supplied value replaces the stored value. - Header updates — An empty mapping (
{}) clears headers. A***entry incdp_connect_headerskeeps the stored value for that key. Values in publicextra_http_headersupdates, including***, are literal. Compatibility change:cdp_connect_headers: nullnow clears headers. Omit the field to keep them. - Copilot settings — Copilot inherits omitted workflow settings and resolves stored values at Accept. Model-facing workflow documents retain root
extra_http_headersandcdp_connect_headersheader names, with every value replaced by***. They omit empty header maps, TOTP identifiers, TOTP verification URLs, webhook URLs, and non-geo proxy mappings entirely. Valid geo targets, including city, subdivision, andisISP, remain visible. Block settings retain their existing document behavior. - URL diagnostics — URL validation errors identify the field and failure class without returning the URL or host. Run-request callback errors identify the callback field. Copilot removes TOTP values from runtime diagnostics sent to the model. Customer runtime diagnostics remain available.
Improvements
- Credential Source Status — Credential settings now show the credential source and keep service account tokens out of the interface.
- Copilot Run Context — Copilot now receives challenge details, recovery options, every result row, and clearer browser error context for better workflow repair.
- Protected Form Submissions — Supported job application forms now wait for challenge completion before the submit continues, reducing duplicate submissions.
- Clearer Run Timeline Statuses — Run timelines now show verified clicks as completed and provide labels for every status glyph.
- Wider Past Runs Range — The agent page now shows the last 365 days by default, so older runs appear without extra filter changes.
- Agent List Search Persists — Agent list searches now survive back navigation, reloads, and shared links.
- Sidebar Links Stay Visible — Resource links now appear on every page with a sidebar, so you can reach help without leaving the page.
Bug fixes
- Fixed dropdown selections failing when the selected label differed from the option text.
New features
- Interactive Questions in Copilot — Copilot can now ask questions with choices or text fields and resume after you answer. (#8447)
- Web Search for Copilot and Code Blocks — Copilot and code blocks can now search the web directly, so workflows can collect results without scraping search pages. (#8464)
- Schema-Directed Parquet Exports — Workflows can now export extracted data to typed Parquet files from a standalone or extraction block. (#8368, #8397)
- Login Blocks from Browser Recordings — Browser recordings now create login blocks from saved credentials and retain native select choices.
- In-Chat Credential Requests — Copilot can now ask for missing credentials in chat and continue after you provide or skip them. (#8420)
- Google Sheets Connections in Chat — Copilot can now connect Google Sheets accounts in chat and resolve unambiguous saved connection names. (#8412)
Improvements
- Embedded-Page Support — Copilot and secure CodeBlocks now keep selected frame data current across navigation, popups, and detached frames. (#8341, #8361)
- Clearer Copilot Failure Diagnosis — Fix with Copilot now sends run details directly; repairs preserve drafts, outputs, page values, selectors, and safety details. (#8343, #8350, #8413)
- More Reliable Browser Recovery — Browser sessions now recover from slow liveness checks, crashed tabs, and stalled screenshot captures. (#8401, #8463)
Bug fixes
New features
- Guided Onboarding for New Users — Eligible new users in enabled organizations get a guided onboarding flow with a working example, progress, and tailored next steps.
- Flexible Credential Setup — Credential workflows now support passwordless entries, 1Password login items, custom one-time-code fields, saved OTP bindings, and separate rotating browser state. (#8317)
- Save Credentials from Record Browser — Record Browser drafts now redact secrets and let users save the matching credential type. (#8307, #8323)
- Failure-Focused Analytics — For organizations with analytics enabled, analytics reports now show failures, selected statuses, action totals, and relevant credit metrics. (#8304)
Improvements
- Public API and MCP Reliability — API and MCP clients now get discoverable metadata, rate-limit guidance, retry responses, and clearer validation errors. (#8283)
- Faster Live Browser Sessions — Live browser sessions now announce loading status, detect dead connections, and recover from temporary stalls. Vendor CDP live views send mouse moves at native rates.
- Copilot Repair Evidence — Copilot now keeps browser evidence, records code failures, and waits for progressing runs before reporting repair results. (#8325)
- Copilot Activity and Responses — Copilot now shows activity, formatted responses, and terminal results in one readable view.
- Google Connected Accounts — Google workflow blocks and Copilot now accept a connected account’s name or email.
- Workflow Inputs at the Start — Workflow editors now show declared inputs beside the starting block.
Bug fixes
- Fixed workflow runs failing to commit downloads, custom selections, dates, and typeahead values. (#8328)
- Fixed run views and workflow editors showing blank pages, wrong live states, hidden outputs, false unsaved-change warnings, or unavailable runs.
New features
- MCP Access and Run History — Hosted MCP clients can now renew access, complete OAuth callbacks, and include child runs in workflow history.
- Human-Approval Steps in Copilot — Copilot can now add a human-interaction step and report paused runs as pauses.
Improvements
- Faster Copilot Block Testing — Copilot can now edit one block and rerun it from the browser state that produced the recorded page.
- Grounded Copilot Page Evidence — Copilot now uses page evidence from the run browser, including screenshots, links, overlays, and hidden authentication options, to guide repairs. (#8239)
- Clearer Copilot Reviews and Failures — Copilot now shows accurate review state and preserves useful failure details for faster repairs.
- More Reliable Sign-In and OTP Handling — Sign-in flows now wait for forms, verify fields, submit fresh codes, recover connected accounts, and report delivery problems clearly.
- Reliable Workflow Files — Workflow files now stay linked to their run, handle download interstitials, and report aborted transfers or unsupported formats clearly.
- More Accurate Form Interactions — Form interactions now verify live controls, clear retry fields, match values accurately, prevent false submission reports, and avoid wrong selectors.
- Past Runs Filters Persist — Past Runs search and status filters now stay in the URL, so back navigation and reload restore the same list.
Bug fixes
- Fixed lost browser recordings, dark live views, and raw errors after browser or MCP connections close.
- Fixed invalid browser session and workflow inputs returning misleading responses or leaving runs without terminal states.
- Fixed billing flows that could start duplicate checkouts or direct exhausted-credit users to unavailable recovery options.
New features
- Focused MCP Connections — MCP now offers focused tool sets for local and hosted connections and clear expired-session guidance.
- Structured MCP Output and Page Reading — The
leanandallMCP tool sets gain an optional final-output validation tool. An opt-in reader tool reads large pages in bounded chunks. - Custom SMTP Delivery — Send Email blocks can send messages through a customer’s SMTP server.
- Full-Page Live Control — Persistent live-browser sessions now keep the full page reachable during takeover.
Improvements
- Copilot Email Verification — Copilot now uses saved credentials or connected inboxes for email verification. (#8200)
- Reliable Download Workflows — Download flows accept locators, recover files, expose the helper early, and confirm registered files.
- Copilot Workflow Editing — Copilot preserves browser context, adds blocks incrementally, retains existing edits, and suggests text-based selectors. (#8205, #8206)
- Copilot Test Run Guidance — Copilot now limits credential approvals and follows test runs with page and challenge evidence. (#8210, #8204, #8214)
- Workflow Studio Preview: Clearer Activity — The opt-in Studio preview retains pane choices, shows terminal browser states, and makes Copilot progress easier to follow.
- Simpler Workflow Integrations — Google integrations need fewer permissions and use native blocks; workflow setup supports optional credentials, parser choices, and root Drive files. (#8202)
Bug fixes
- Fixed incorrect toggles, lost outputs, unclear run errors, stale browser actions, credential or dropdown failures, and repeated Copilot requests.
New features
- Credential Run Controls — You can serialize runs sharing a credential while fresh profiles omit old login files and saved profiles retain recent sign-ins. (#7995, #8001, #8097)
- MCP Local Browser Access — MCP can now control an existing local browser, and its bridge starts serving requests promptly. (#8070, #8175)
- External Browser Live View — You can view and navigate externally hosted browser sessions, and use keyboard editing in live view. (#8101, #8114, #8143)
- Modern Code Blocks — Code blocks use the editor shape, open shared-session tabs, recover transient startup timeouts, and emit typed errors for downstream routing. (#8006, #8022, #8033, #8034, #8055, #8163)
- Bulk Agent Export — Bulk exports now download as one ZIP with a separate file for each agent. (#8105)
Improvements
- Deterministic Custom-Select Inputs — Typed entries now use exact matching when possible, reducing model fallback and incorrect success reports. (#7996)
- Copilot Credential Assistance — Copilot now finds credential IDs, lets you choose credentials, shows automatic selections, and starts from a credential’s saved login page. (#8026, #8013, #8065, #8064, #8061, #8094, #8099)
- Reliable Email Verification Retrieval — Email OTP and sign-in-link lookup now preserves exact values, reads externalized message parts, and respects the expected code type. (#8028, #8062, #8091)
- Reliable Browser Session Artifacts — Downloads now complete more reliably, and recordings remain available after browser-session teardown. (#8007, #8021, #8019, #8115, #8125, #8167)
- Copilot Workflow Editing — Copilot now preserves workflow code and data, shows plain-language run steps, supports longer repairs, and returns successful results. (#8058, #8104, #8106, #8142, #8122, #8166)
Bug fixes
- Fixed Copilot turns stopping valid actions, resending queued prompts, or showing empty expansion controls. (#8025, #8060, #8089, #8137)
- Fixed browser sessions failing after stale connections, leaving released occupancy behind, or reporting false navigation, closure, expiry, or final-status errors. (#7997, #8043, #8164)
New features
- Expanded OTP Recovery — OTP handling now preserves raw content, detects code types, reads connected email accounts, and supports magic links in credential view. (#7849, #7859, #7866, #7886, #7901, #7908)
- Weekly Analytics Email Digest — Eligible organizations with the digest enabled receive a weekly email summary with recent activity, credit use, and lowest-success workflows.
- Scoped Browser Session Access — Browser sessions can start at a requested URL. In supported regions, enabled first-party sessions use scoped router addresses for live browser viewing. (#7915)
Improvements
- Job Application Form Handling — Supported job application forms now handle rich fields, rejected values, and live credential pages, with optional visual secret masking. (#7771, #7773, #7777, #7955, #7966, #7967)
- Shorter Browser Session Lifetime — Browser sessions now have a maximum lifetime of 4 hours instead of 24. Longer timeouts are clamped, and Enterprise accounts can request longer sessions. Recordings, downloads, and live-stream paste are more reliable. (#7772, #7774, #7804, #7838, #7883, #7902, #7916, #7929, #7930, #7944, #7979)
- Copilot Authoring Reliability — Copilot verifies outputs, preserves drafts, uses connected accounts, acts on pages, reports activity and failures, and keeps repairing through stalls. (#7779, #7791, #7792, #7794, #7796, #7802, #7806, #7845, #7869, #7871, #7873, #7878, #7885, #7911, #7913, #7914, #7934, #7972)
- Safer MCP and Browser Requests — MCP tools now handle arguments and page waits more reliably while browser and webhook requests validate destinations and headers. (#7765, #7782, #7932, #7941, #7962, #7976, #7982)
- Clearer Billing Usage — The billing page now shows OTP usage.
- Google Sheets Workflow Fields — Spreadsheet and sheet pickers now show compact destinations and support template expressions for dynamic targets. (#7840, #7875)
- Larger File Workflows — File workflows accept larger storage uploads, return ZIP file lists, and process up to 1,000 loop iterations. (#7793, #7899, #7907)
Bug fixes
- Fixed CAPTCHA actions showing false completion, missing checkbox frames, or losing settings during cached replay. (#7787)
- Fixed workflow runs reusing stale outputs or losing status, webhook delivery, browser pages, timeline data, code-block execution, or run-pane access. (#7775, #7809, #7827, #7841, #7842, #7864, #7890, #7903, #7953, #7960, #7978)
New features
- Password-Free Login and CAPTCHA Handling — Copilot can use email sign-in codes or links and solve CAPTCHA challenges during login exploration. (#7727, #7755, #7756, #7761)
- Live Action Progress During Test Runs — Code-block test runs stream recorded browser actions while they run. The Copilot preview shows these actions live for enabled organizations. (#7608, #7609, #7610)
- MCP Setup and Session Visibility — MCP now offers in-app setup, live timelines, clear argument errors, code-only guidance, and profile cleanup. (#7587, #7574, #7619, #7693, #7577, #7580)
- Workflow Studio Preview: Run Navigation — The opt-in Studio preview makes past runs easier to find with linked titles, shorter URLs, clear details, and branch navigation. (#7602, #7604, #7607, #7642, #7643, #7644, #7646, #7667, #7758)
- Agent Filter for Run History — Run History now filters runs by agent while keeping server-side search and pagination. (#7618)
- Claude Opus 5 Model — Claude Opus 5 is now available as a selectable model for agent runs for Enterprise accounts on Skyvern Cloud. (#7741)
- Paused Status in the Public API — The public run status now includes paused runs, so status reads and webhooks work during verification waits. (#7576)
Improvements
- More Reliable Copilot Workflow Authoring — Copilot now completes more requests, repairs missing workflow steps, handles credentials safely, and reports outcomes honestly. (#7563, #7565, #7567, #7613, #7628, #7629, #7630, #7632, #7648, #7649, #7656, #7658, #7673, #7686, #7688, #7689, #7695, #7699, #7701, #7703, #7711, #7713, #7714, #7715, #7716, #7719, #7734)
- Job Application Form Reliability — Supported job application forms now handle more field widgets, recover rewritten inputs, and confirm successful submissions. (#7570, #7585)
- More Reliable Remote Browser Startup — Remote browser startup now retries transient connections, reports rate limits clearly, uses fallback browsers, and supports custom proxies.
- CodeBlock Authoring and Output Improvements — CodeBlock workflows now show outputs and safe errors, support page evaluation, wire inputs, carry MCP metadata, and make JSON searchable. (#7578, #7586, #7670, #7697, #7730, #7733, #7754)
Bug fixes
New features
- External File Storage Targets — Download blocks can now send files to external storage, and large Google Drive uploads resume after interruptions. (#7444, #7373, #7391)
- Record and Synthesize Browser Sessions — MCP sessions can now record safe browser actions and turn them directly into replayable code blocks. (#7498, #7522)
- Credential Help in Copilot — The Copilot preview shows a credential card with Connect and Skip actions for enabled organizations. The card appears when a build needs access. (#7418)
- Browser Profile Refresh and Usage Checks — Browser Profiles now show dependencies, support refresh, and warn before deleting a profile in use. (#7548, #7550)
Improvements
- Grounded Copilot Workflows — Copilot now preserves captured steps and completion checks, recognizes values by format, reuses inputs, and reports verified downloads. (#7374, #7349, #7381, #7375, #7397, #7439, #7438, #7441, #7455)
- Safer Copilot Authoring — Copilot now checks goal reach, code validity, and completion conditions before saving, then gives clear retry guidance. (#7419, #7429, #7466, #7507, #7556, #7442, #7512, #7336, #7527, #7549, #7557)
- More Reliable Browser Actions — Browser runs now verify form controls and wait for protected pages. (#7334, #7406, #7454, #7480, #7483, #7474)
- Reliable Artifacts and Downloads — Artifact views refresh expired links, while downloads reject login pages and missing profiles start faster. (#7459, #7484, #7411, #7350)
- MCP Workflow and Browser Tools — MCP now gives clearer guidance, valid snippets, file-based edits for large workflows, and selected iframe content. (#7333, #7460, #7422, #7433)
- Analytics and Run Health — For organizations with analytics enabled, analytics now shows accurate credit usage and run outcomes. Runs retry transient failures, and run lists surface self-healing risk. (#7493, #7494, #7523, #7526, #7541)
- Workflow Studio Preview: Usability — The opt-in Studio preview now follows themes, shows changes, links run review to blocks, and restores run inputs and outputs. (#7408, #7417, #7513, #7409, #7495, #7388, #7423, #7426, #7543)
Bug fixes
New features
- Workflow Studio Preview: Navigation and Layout — The opt-in Studio preview now searches blocks, expands hidden containers, jumps through long workflows, and remembers panel layouts. (#7103, #7104, #7107, #7112, #7118, #7201, #7270)
- Workflow File Tools — Workflow file tools can now split PDFs, select upload files, deliver files to SFTP, and use sandbox scratch files. (#7258, #7268, #7301)
- Organization OAuth Setup — Organization-level OAuth setup is available for enabled organizations, and connections that need reconnection are clearly marked after settings changes. (#7283, #7288)
- Email Inbox Workflows — Email Inbox workflows can read matching Gmail or Outlook messages for downstream processing. (#7303, #7309)
- Run History and Result Links — Run history now filters by run type, while the Studio preview and MCP expose available outputs and recording links. (#7251, #7269, #7307)
Improvements
- Copilot Build Progress and Recovery — Copilot and agent runs now show live progress and recover from build, login, and model-response failures. (#7097, #7098, #7099, #7100, #7120, #7149, #7155, #7162, #7216, #7246, #7275, #7277, #7308, #7313, #7314, #7315, #7327)
- Analytics Console Clarity — For organizations with analytics enabled, Analytics now uses shared filters, clearer charts, accurate workflow breakdowns, and stable custom date ranges. (#7136, #7209)
- MCP Browser Reliability — MCP now connects reliably, guides keys to setup, accepts query aliases, protects inputs, retains references, and supports direct dropdown actions. (#7165, #7166, #7203, #7206, #7220, #7250, #7285)
- Browser Automation Reliability — Browser automation now handles captchas, captures client-created files, preserves download names, and avoids extra recordings. (#7102, #7111, #7123, #7153, #7156, #7172, #7192)
Bug fixes
- Fixed credential setup so editable accounts, saved instructions, reused profile names, test cleanup, and token errors work correctly. (#7122, #7157, #7160, #7184, #7219)
- Fixed template run details failing to load, loading slowly, or omitting the Stop control in the Studio preview. (#7105, #7265, #7330)
- Fixed run inputs hiding download instructions, corrupting formatted card numbers, and mishandling workflow files or malformed definitions. (#7202, #7260, #7264, #7305, #7306)
Workflow Studio preview panes, faster enrichment and live recording, code-first authoring, file parsing, credential and secret handling, code-block self-heal, extraction accuracy, bug fixes
New features
- Workflow Studio Preview: Side-by-Side Panes — The opt-in Workflow Studio preview now offers side-by-side panes, responsive layouts, and clearer run views. (#7007, #7008, #7009, #7010, #7013, #7021, #7022, #7024, #7025, #7043, #7044, #7045, #7046, #7047, #7056)
- Code-First Workflow Authoring — Teams can edit workflows as YAML or use MCP tools to lint, synthesize, validate, and save code-only workflows. (#6997, #6998, #7017, #7030, #7065)
- ZIP Files in File Parser — File Parser blocks now process ZIP archives, so workflows can extract data from bundled documents. (#7049)
- Credential and Session Reuse — Workflows can rotate login credentials across runs and save browser state in managed profiles. (#7032, #7053)
- Secret Values Across Blocks — Workflows can pass selected HTTP response values as secrets between downstream blocks. Enabled self-hosted deployments can store credentials locally. (#7054)
Improvements
- Block Type Shown in Workflow Studio Header — The block detail panel in Workflow Studio now displays the block type in the header, making it easier to identify what kind of block you’re editing at a glance. (#6956)
- Improved Extraction Accuracy for Absent Data — The agent now better handles cases where requested output is genuinely absent from the page, reducing false positives in extraction results. (#6957)
- Data-Only Validation Skips Page Evidence — Validation steps that check only extracted data no longer capture page screenshots when they aren’t needed, reducing unnecessary work during runs. (#6960)
- Copilot Block Runs Routed to UI Worker — Copilot block runs are now dispatched to a dedicated UI worker, keeping the editor responsive during live automation. (#6962)
- Clearer Invalid JSON Errors in Workflow Editor — The workflow editor now surfaces a descriptive error when a parameter contains invalid JSON, making it faster to diagnose and fix configuration mistakes. (#6963)
- “Ignore System Prompt” Moved to Advanced Settings — The “Ignore system prompt” option has been relocated to the advanced settings panel, reducing clutter in the default block configuration view. (#6964)
- Empty Extra HTTP Headers Hidden — Extra HTTP header fields that are empty no longer appear in block configurations, reducing visual noise in the workflow editor. (#6966)
- Block Selection Persisted in URL — The selected block in Workflow Studio is now encoded in the URL, so your selection is preserved on reload and can be shared via link. (#6969)
- Code-Block Self-Heal Minimum Retry Floor — The code-block self-healing loop now enforces a minimum number of repair attempts before giving up, preventing premature failures on recoverable errors. (#6971)
- Faster Live-Draft Enrichment — Workflow recording live-draft enrichment now uses a faster model and a shorter debounce interval, making draft suggestions appear noticeably sooner while recording. (#6973)
- Responsive Browser Recording — Browser recording now shows editable live drafts and stays responsive across long sessions, repeat recordings, and browser transport changes. (#6995, #6999, #7002, #7004, #7005, #7036, #7048)
Bug fixes
- Fixed syntax highlighting disappearing when viewing large webhook payloads. (#6967)
- Fixed LLM blocks not enforcing their configured output schema, which could allow malformed data to propagate to downstream blocks. (#6965)
- Fixed Copilot repairs and builds that failed or stopped early when output evidence, action history, or prior outputs were incomplete. (#6993, #6994, #6996, #7000, #7014, #7015, #7031, #7040, #7055, #7069, #7081, #7086, #7087, #7088, #7090, #7091)
- Fixed failures in workflow navigation, remote browser cleanup, and task listings. (#7001, #7006, #7042, #7084, #7085)
Workflow Studio preview, new models, label management, Gmail OTP, Copilot improvements, analytics filters, bug fixes
New features
- Gmail-Backed OTP Lookup — Agents can now retrieve one-time passwords from Gmail via OAuth, enabling automated login flows that depend on email-delivered verification codes. (#6769)
- Copilot Chat History per Workflow — The Copilot now retains its full conversation history scoped to each individual workflow, so prior context and decisions are preserved as you continue building. (#6767)
- Code-First Editor Polish — Code-first code blocks now display block labels, correctly render extraction output, and include a Build-vs-Code toggle for switching between visual and code editing modes. (#6771)
- Workflow Studio (Preview) — A redesigned workflow editor with a refreshed build canvas, integrated browser view, and run viewer is now available as an opt-in preview, giving teams an early look at the next-generation authoring experience. (#6782)
- DeepSeek and Xiaomi MiMo Models — DeepSeek and Xiaomi MiMo are now available in the model dropdown for agent runs. Gemini 2.5 Pro and 2.5 Flash have been deprecated. (#6831)
- Label Management Settings — Labels can now be renamed, recolored, and soft-deleted directly from the Settings page, making it easier to maintain a clean tag taxonomy. (#6830)
- Browser Profiles in MCP Sessions — Browser sessions initiated via MCP can now be associated with a browser profile, enabling credential and state reuse across MCP-driven runs. (#6798)
- Workflow-Tag Filter on Analytics — The analytics dashboard now supports filtering runs and metrics by workflow tag. (#6780)
Improvements
- Copilot Repair Shows Quiet Progress — The Copilot’s code-seam repair steps now appear as compact progress indicators instead of repeated chat messages, keeping the conversation readable during long repairs. (#6746)
- Collapsible Sidebar with Avatar — The left sidebar now features an edge toggle, a closed-state user avatar, and a smooth open/close animation for a more polished navigation experience. (#6754)
- Structured Output Verified Before Completion — Runs that produce structured data now verify the output satisfies the configured schema before marking the run as successfully complete. (#6751)
- MCP Workflow Overlap Settings Preserved — Overlap settings configured via MCP are now retained correctly when the workflow is subsequently edited, preventing unintentional resets. (#6748)
- Browser Profile List Sorted Newest First — The browser profile picker now orders profiles with the most recently created at the top, making recent profiles easier to find and select. (#6739)
- Remote Browser Sessions Capped at Four Hours — Remote browser sessions now have a maximum duration of four hours, preventing orphaned sessions from consuming resources indefinitely.
- Debug Sessions Stay Open After Runs — Browser sessions launched from the workflow editor in debug mode now remain alive after a run completes, so you can inspect the final page state without the session closing immediately. (#6832)
- GPT-5 Mini via OpenRouter — GPT-5 Mini is now available as a selectable model via OpenRouter. (#6816)
- Extraction and AI Steps in Code-Block View — Extraction and AI steps are now surfaced in the code-block run view, filling a gap in execution visibility. (#6823)
- Color Palette for Grouped Workflow Tags — Tags that belong to a group are now displayed with a curated color palette, making it easier to distinguish tag groups at a glance. (#6801)
- Improved Authenticator 2FA Credential Setup — The credential setup flow for authenticator-based 2FA has been improved with clearer guidance and a smoother configuration experience. (#6783)
- Copilot Data-Write Blocks Default to Safe Failure Mode — New Copilot-generated data-write blocks now default to
continue_on_failure=false, preventing downstream blocks from running on bad data. (#6779) - End-State Screenshots for All Open Tabs — Task runs now capture end-state screenshots for every open tab at completion, giving a full snapshot of the browser state. (#6777)
- Partial Deliverable on Budget Exhaustion — Tasks that hit their credit budget now return the best-effort deliverable collected so far instead of failing without output. (#6778)
- Planner Supports EXTRACT_INFORMATION and GOTO_URL — The task planner can now emit EXTRACT_INFORMATION and GOTO_URL actions, expanding the set of strategies available during automated planning. (#6800)
- Copilot Completion Recognition Improvements — The Copilot completion recognizer is now more authoritative and evidence-backed, with stable criteria counts and more reliable recognition across runs. (#6789)
- Copilot Code Repair Loop Reliability — The Copilot code authoring repair loop is now more reliable, reducing cases where stuck repairs would stall progress. (#6792)
- Workflow Studio Browser and Panel Improvements — The Workflow Studio editor now supports take-control and viewport centering for the live browser, with alignment and toolbar polish for the run/editor panels. (#6810)
- Analytics Tag Filters Accept Flexible Input Shapes — Analytics tag filters now accept labels, groups, or
group:labelpairs in all three supported filter shapes. (#6802) - Run ID Search Fallback on Runs List — The runs list now falls back to searching by run ID when no parameter-value match is found, making it easier to locate specific runs. (#6788)
Bug fixes
- Fixed login credentials not being reattached after editing a workflow via MCP, which could cause credential-gated runs to fail. (#6761)
- Fixed conditional routing logic not being preserved when loading a workflow stored in the v1 format. (#6760)
- Fixed the active run context being lost when switching between organizations, requiring a page refresh to restore the correct run view. (#6753)
- Fixed browser connection settings defined in a TaskV2 block not being forwarded to child task runs. (#6750)
- Fixed telephone number inputs dropping digits when bare NANP-format phone numbers were typed. (#6826)
- Fixed the workflow builder freezing due to an infinite FlowRenderer dimension/layout feedback loop. (#6824)
- Fixed the code editor crashing with a stack overflow when opened on large or deeply-nested JSON. (#6817)
- Fixed the app failing to load after a deploy due to stale webpack chunks; the app now auto-recovers via a cache-busting reload. (#6818)
- Fixed runs returning an incorrect terminal outcome when the extraction schema had been modified after the initial recording. (#6815)
- Fixed MCP workflow edits dropping non-credential parameter-to-block links. (#6813)
- Fixed address bar paste not working in the VNC streaming view. (#6812)
- Fixed newly added workflow blocks not receiving focus in the editor. (#6809)
- Fixed Copilot code-block file downloads being written to the wrong directory. (#6797)
- Fixed parameter-value search not appearing in the v2 runs list. (#6796)
- Fixed extra actions firing after selecting a combobox option during text entry. (#6787)
- Fixed shadow DOM text nodes being missed during element text extraction. (#6791)
- Fixed persistent browser sessions incorrectly reporting a cached session as live when it had become stale.
Code-first code block editor, analytics drill-down, copilot repair loop, OTP fill, uncapped runtimes, voice input fixes
New features
- OTP Fill at Runtime in Code-Block Sandbox — Code blocks can now fill OTP fields using your stored TOTP credentials at runtime, enabling fully automated two-factor authentication flows without manual input. (#6571)
- Uncapped Workflow Runtime — Workflows can now be configured to run without a maximum time limit, enabling long-running automations that previously timed out. (#6567)
- Code-First Code Block Editor — Code blocks now feature a dedicated code editor with syntax highlighting, Jinja parameter hints, and a run timeline showing each action taken, giving developers full visibility and control over code-driven automations. (#6604)
- Period-Aware Analytics Drill-Down — The agent analytics drill-down panel now respects the selected time period (7/30/90/365 days or custom range) and shows per-agent average duration, credits per run, and period-over-period trends.
- Agent Directory Tree — The agents and folders list has been redesigned as a navigable directory tree, making it easier to browse and organize agents across nested folder hierarchies. (#6631)
- GCP Self-Host: Storage and Vault — Self-hosted deployments on Google Cloud Platform now support GCS artifact storage and GCP Secret Manager vault, enabling full GCP-native deployments. (#6632)
- 1Password Credential Picker — Credential inputs now include an item picker for 1Password vaults, letting you browse and select vault items directly during credential setup. (#6651)
- Typed Extraction Schema for Copilot Code Blocks — Copilot-generated code blocks now support a typed data extraction schema (matching
data_schemaparity), enabling structured data output from code-block automations. (#6674) - SDK-Backed Remote Browser Provider — Agents can now connect to external browser providers via SDK, enabling tighter integration with third-party remote browser services. (#6687)
- Bitwarden Credential Integration — Saved credentials can now be sourced directly from Bitwarden vaults, bringing Bitwarden to full UI and functional parity with other supported credential providers. (#6729)
Improvements
- Smarter Loop Planning for Repeated Work — The task planner now prefers loop-based tasks when it detects repeated same-shaped operations, improving throughput on multi-item automation runs. (#6575)
- Agent Auto-Acts on Stalled Pages — When the agent re-evaluates a page that hasn’t changed, it now applies steering heuristics and acts immediately instead of stalling. (#6577)
- Final Answer Surfaced Directly — The task agent now returns the final deliverable directly instead of spinning through redundant synthesis steps, reducing latency on completed tasks. (#6574)
- Copilot Stops Stuck Repair Loops — The Workflow Copilot now detects when consecutive repairs make no verified progress, halts after three attempts, and presents an honest escalation with the draft preserved for review. (#6608)
- Voice Input Microphone Permission — The browser microphone permission prompt now appears correctly when using voice dictation in the Prompt Box and Workflow Copilot. (#6601)
page.evaluateRecorded in Code Block Timeline — JavaScriptpage.evaluate(...)calls made inside code blocks are now recorded as actions in the run timeline, filling a gap in execution visibility. (#6600)- Active Run Shown on Browser Session Page — Browser session detail pages now display which workflow run currently occupies the session, making it easier to understand session usage. (#6691)
- Distinct Visual for Terminate Actions — Terminate actions now appear with a distinct amber triangle icon in the run timeline, making it easy to spot where a run was stopped. (#6693)
- PDF Field Mapping by Positional Labels — PDF form filling now uses positional field labels to improve accuracy when mapping data to form fields with ambiguous names. (#6663)
- MiMo-V2.5 Model Support via OpenRouter — Added Xiaomi MiMo-V2.5 as a selectable LLM via OpenRouter for agent runs. (#6636)
- Settings “API Keys” Renamed to “General” — The “API Keys” section in Settings has been renamed to “General” to better reflect its full scope of configuration options. (#6642)
- Run URLs Surfaced in MCP and CLI — The agent app URL and session recording URLs are now included in run responses returned via MCP and the CLI, making it easier to link directly to a live run. (#6719)
- Email OTP Credentials Supported in Copilot — The Copilot can now use stored email OTP credentials when generating and validating code-block automations that require email-based authentication. (#6716)
- Auto-Fill Default Google Account for Sheets Blocks — Google Sheets blocks now automatically select the default linked Google account, reducing manual setup when configuring Sheets integrations. (#6715)
- Copilot Output Accurately Reflects Verified Results — The Copilot no longer under-reports successfully verified output, ensuring the run summary accurately reflects what was accomplished. (#6710)
- Runs Halt Reliably on Terminal Blockers — When the agent encounters strong evidence of an unresolvable blocker (such as a persistent challenge page), it now stops immediately rather than continuing to attempt further actions. (#6714)
- Workflow Save Performance Improved — Cache invalidation on workflow save is now batched, eliminating a per-block N+1 query pattern that slowed saves for large workflows. (#6718)
Bug fixes
- Fixed the Run Inputs panel collapsing unexpectedly when filtering Run History. (#6573)
- Fixed the agent incorrectly retrying click fallbacks after an action dispatch timed out, causing unintended interactions. (#6569)
- Fixed session recordings and page interpretation not resuming correctly after navigation, causing missed recording segments. (#6568)
- Fixed code block runs not generating a recording entry, causing the Recording tab to appear empty. (#6607)
- Fixed code block action screenshots not appearing on the run Overview page. (#6605)
- Fixed long usernames or email addresses hiding the sidebar collapse button on desktop.
- Fixed the FileParserBlock crashing instead of surfacing a clear error when it encountered an unparseable file. (#6657)
- Fixed the Inputs field missing from the Code view of code-first code blocks. (#6633)
- Fixed Copilot-synthesized download blocks not downloading the file during code replay. (#6637, #6638)
- Fixed multi-step login flows not advancing when credentials were filled into JavaScript-gated fields. (#6650)
- Fixed an infinite update loop on the workflow build page when certain block configurations were active. (#6660)
- Fixed adopted-session file downloads not being available as run artifacts after the session ended. (#6679)
- Fixed the run timeline showing action and step rows for non-code blocks when they should only appear for code blocks. (#6675)
- Fixed workflow run actions not appearing in the timeline for all block types.
- Fixed workflow runs not stopping when a configured time limit was exceeded. (#6677)
- Fixed the analytics plan-tier badge showing the incorrect tier for some organizations.
- Fixed credits not being recorded when data was captured inside a Navigate block in task runs. (#6698)
- Fixed Copilot directory mode synthesizing incorrect code bindings. (#6702)
- Fixed FileUpload blocks silently skipping the upload when no file was provided, instead of recording a clean no-op and continuing. (#6728)
- Fixed the product tour not launching when triggered from the “Take a tour” menu option. (#6727)
- Fixed scanned PDF files being parsed as a single large page rather than processed page-by-page, which caused extraction to miss content on multi-page documents. (#6722)
- Fixed the block library drawer overflowing its container bounds in the workflow builder. (#6721)
Multi-tab agent control, Gemini 3.5 Flash, GCP Vault, bulk agent actions, credential clears, code-block credentials
New features
- Copilot Ask/Build/Code Mode Routing — The Copilot now routes each request based on the active mode: Ask returns answers without modifying the workflow, Build creates or updates workflows, and Code generates executable code blocks. (#6459)
- GCS Storage Backend — Added Google Cloud Storage as an artifact storage backend for self-hosted deployments. (#6442)
- Key-Optional Labels and Flexible Tag Filters — Tags now support key-optional labels, and tag filter queries accept more flexible formats for richer agent organization. (#6458)
- Self-Serve Editor Onboarding — New users in the workflow editor now see a get-started modal with intent routing and template cards, a 4-stop guided product tour, and smart empty states on the dashboard and runs pages to accelerate time-to-first-automation. (#6475)
- Multi-Tab Browser Control — The autonomous agent loop can now open and interact with multiple browser tabs simultaneously, enabling workflows that span across new tab links and pop-ups. (#6507)
- Gemini 3.5 Flash Model Support — Added Google Gemini 3.5 Flash as a selectable LLM for agent runs, providing a faster inference option. (#6505)
- GCP Secret Manager Vault Adapter — Added Google Cloud Secret Manager as a supported vault backend for storing and retrieving secrets in self-hosted deployments. (#6488)
- Bulk Actions on Agents List — You can now select multiple agents at once and apply bulk actions directly from the agents list page. (#6485)
- Self-Service Credential Clears — Saved credentials can now be individually cleared from the credentials settings page without deleting and recreating them. (#6480)
- Per-Run Session Recordings as Artifacts — Recordings are now clipped to each run’s time window and stored as downloadable run-scoped artifacts. (#6479)
- Credential-Aware Scouting in Code-Block Mode — The Copilot now uses saved credentials while scouting pages for code-block generation and produces credential-typed code that fills login fields using your stored credentials. (#6521)
- Claude Fable 5 Model Support — Added Claude Fable 5 as a selectable model for agent runs. (#6539)
- Live Browser Recording with Instant Placeholders — Session recordings now commit navigation and wait blocks immediately as placeholders during live recording, with support for capture pause and draft interpretation. (#6561)
- Act-and-Observe Copilot Scouting — The Copilot’s page scouting now uses act-and-observe interactions that return bounded page evidence alongside gathered actions, improving code-block generation accuracy. (#6533)
Improvements
- Deterministic Copilot Code-Block Synthesis — Code blocks generated by the Copilot from scouted page interactions are now deterministic, producing consistent scripts across repeated generations. (#6463)
- Smarter Element Focus in Page Analysis — The agent now drops non-navigable links from its internal element tree, reducing noise and improving targeting accuracy. (#6461)
- Workflow Builder Left Sidebar Preserved — The left sidebar in the workflow builder now stays open when you switch between blocks, keeping your panel state between edits. (#6453)
- TOTP Prioritized Over Webhook for OTP — Stored TOTP secrets are now tried before webhook polling when retrieving one-time passwords, speeding up authentication flows that use stored credentials. (#6451)
- Improved Caching for Click-Reached Blocks — Scout interactions for blocks reached via click are now automatically credited toward the cached script, increasing cache hit rates. (#6450)
- Accessible Status Badges — Status badges across the product have been updated with improved contrast and consistent pill styling for better accessibility. (#6444)
- Block Title Editable from Settings Sidebar — Workflow block titles can now be edited from the right-hand settings sidebar, without needing to use the inline editor on the canvas. (#6443)
- API Reference Split into Agents and Runs Sections — The public API reference is now organized into separate Agents and Runs sections for easier navigation. (#6448)
- Label-First Tag Editor and Flexible Tag Filter UI — The tag editor now accepts standalone labels or
group:labelformat in a single input, and the filter bar supports filtering by label, group, or group:label for richer agent organization. (#6470) - 2FA Field Progressive Disclosure in Login Block — Two-factor authentication fields in the login workflow block are now collapsed by default and expanded on demand, reducing visual noise for workflows that don’t use 2FA. (#6490)
- Persistent Browser Session Profile Generation Now Opt-In — Automatic browser profile generation during persistent sessions is now opt-in, giving you explicit control over when profiles are captured. (#6489)
- Icon-Only Status Pills on Mobile Viewports — Status badges now render in a compact icon-only format on small screen widths, preventing layout overflow on mobile devices. (#6481)
- Hardened Webhook Delivery — Workflow webhook delivery is now more reliable, with improved handling of transient failures and prioritized delivery ordering. (#6478)
- Reduced Live Recording Draft Latency — Live session recordings now appear almost instantly with placeholder entries, with final blocks committed after a short debounce window. (#6553)
- Copilot Stops on Terminal Blockers — The Copilot now correctly halts when it reaches a condition that blocks further progress, preventing runaway sessions. (#6535)
- Copilot Chat Shows Run Outcome Verdict — The Copilot chat timeline now reflects the actual run outcome verdict, giving a more accurate view of what the agent accomplished. (#6532)
Bug fixes
- Fixed org-verified discovered credentials not being credited in the Copilot credential gate, which could prevent runs from starting. (#6462)
- Fixed OTP codes from previous workflow runs being picked up and incorrectly used in new runs. (#6455)
- Fixed persistent-session screen recordings not appearing in the run view when the recording timestamp fell outside the run’s time window. (#6454)
- Fixed Copilot inconsistently asking for a URL before running discovery when the user had only named a site without providing one. (#6473)
- Fixed Copilot asking for a URL after already successfully navigating to the requested site during discovery. (#6472)
- Fixed self-hosted deployments rejecting requests with “Invalid credentials” (HTTP 403) when the Docker UI injected an API key that the container’s server incorrectly filtered out.
- Fixed persistent session recordings not being finalized when the video encoding process outlasted the session’s cleanup window. (#6519)
- Fixed the saved credentials status indicator not updating until the page was manually refreshed. (#6517)
- Fixed the workflow run timeline rendering blocks in branch-tree order instead of execution order, producing a confusing step sequence. (#6515)
- Fixed downloaded files inside loop blocks being shared across all iterations instead of scoped to the block that downloaded them. (#6487)
- Fixed session recordings defaulting to an incorrect resolution; recordings now use Playwright’s default viewport size. (#6524)
- Fixed Copilot failing to recognize and use a saved credential when the credential ID was typed with a space or hyphen separator instead of an underscore; malformed IDs are now normalized and resolved correctly. (#6529)
- Fixed Copilot chat showing workflow blocks as completed even when the overall run goal was not demonstrated — the recorded run outcome verdict is now surfaced in the chat narrative. (#6528)
- Fixed Copilot browser sessions incorrectly blocking localhost navigation during page scouting, which could prevent evidence from being gathered from locally-served pages. (#6527)
- Fixed single-element JSON arrays being unwrapped to a scalar when passed as workflow run inputs. (#6560)
- Fixed code blocks silently dropping non-dict return values instead of passing them through to downstream blocks. (#6556)
- Fixed the task agent ignoring the
max_iterationsoverride, causing runs to exceed their configured step limit. (#6544) - Fixed generated workflow parameter keys being duplicated, which caused a hard failure when the workflow was run. (#6543)
- Fixed task runs completing successfully even when a required sub-goal had not been satisfied. (#6546)
- Fixed the credential-clear confirmation dialog being hidden behind other overlays. (#6542)
- Fixed click actions on elements wrapped in disabled containers not being retargeted to the clickable child element. (#6540)
- Fixed Copilot loop detector incorrectly triggering at block dispatch boundaries. (#6550)
New features
-
Workflows are now Agents — We’ve renamed “Workflows” to “Agents” across the product, public API, and docs. The API is now served under
/agents(e.g.POST /v1/run/agents) and accepts anagent_idalias forworkflow_id(samewpid_value), withagent_idechoed in responses. Fully backwards-compatible: existingworkflow_idinputs,/workflowsendpoints, SDK methods, and bookmarked doc URLs all keep working. - Feature Preview Section in Settings — A new “Feature Preview” panel in account settings lets you opt into upcoming features before they are fully released. (#6318)
-
Filter Workflows by Tags — The
GET /workflowsAPI now accepts a?tags=query parameter to filter results by tag. (#6334) - Workflow Schedules in Open Source — Self-hosted Skyvern deployments now support workflow schedules, previously a cloud-only feature. (#6344)
- Claude Opus 4.8 Support — Added Claude Opus 4.8 as a selectable model for both direct Anthropic and AWS Bedrock inference paths. (#6409)
- Automatic Re-save of Expired Credential Sessions — Saved credential sessions that have expired or encountered an error are now automatically re-saved, keeping browser profiles available without manual intervention. (#6435)
- GPT-5-mini Support — Added GPT-5-mini as a selectable LLM for customers using their own OpenAI API keys, providing a faster, lower-cost inference option.
Improvements
- Improved CAPTCHA Solving Reliability — Added a timeout circuit breaker and better timing around CAPTCHA solves on submit-protected pages, reducing failures caused by timing races. (#6349)
- Undo/Redo Moved to Canvas Toolbar — Undo and redo controls are back on the workflow canvas toolbar for easier access. (#6310)
- Better Loading and Error States in Browser Streams — Improved copy and animations for loading and error states across all browser stream views. (#6332)
- Extended Credential Test Timeout — The browser profile credential test now polls for up to 10 minutes, accommodating slower authentication flows that previously timed out. (#6359)
- Code-First Copilot Validation Gates — The code-first Copilot now runs preflight checks, MCP evidence gates, and validation steps before executing generated code blocks, reducing silent failures. (#6410)
- OpenRouter Cost Tracking — LLM token counts and costs are now tracked for runs using OpenRouter models, making per-run spending visible alongside other LLM providers. (#6427)
- CUA and Claude Opus Models Now Enterprise-Only — The CUA browser-control mode and Claude Opus model family are now reserved for enterprise accounts. (#6420)
Bug fixes
- Fixed session cookies not being restored when reusing a saved browser profile. (#6362)
- Fixed TOTP codes not being recovered from webhook response bodies that are not strictly valid JSON. (#6354)
- Fixed webhook responses being truncated at 2KB when routed through the NAT proxy.
- Fixed filtered dropdown options not being selectable in certain search-driven input components. (#6348)
- Fixed block prompts not rendering in the workflow version comparison view. (#6331)
- Fixed agent navigation buttons rendered via CSS
::before/::afterpseudo-content icons being missed when the icon element’s bounding rect was zero due to out-of-flow positioning. (#6385) - Fixed credential edit form values and in-progress test state being cleared when the credentials list refreshed in the background during a browser profile test. (#6404)
- Fixed workflow canvas layout not updating correctly when a container block is collapsed. (#6436)
- Fixed session cookie capture not persisting correctly when saving browser credential sessions. (#6434)
- Fixed persistent browser session final status not being reconciled when a session close races a timeout cleanup.
- Fixed shadow DOM inputs inside closed dropdown components incorrectly being forced visible. (#6433)
New features
- Redesigned Sidebar Navigation — The app sidebar has been redesigned with collapsible groups, improved visual hierarchy, and a cleaner layout. (#6211)
- Workflow Run Retry — Workflow runs can now be retried directly from the UI or API without creating a new run from scratch. (#6195)
- New Workflow Run Page Layout — The workflow run detail page has been split into a compact timeline and a detail panel for faster navigation and better readability. (#6214)
- Copilot Turn Narrative — The workflow copilot now displays a summary bubble after each turn, giving you a clear audit trail of what the agent did and decided. (#6231)
- Block Completion Progress on Run Page — The workflow run page now shows an N/M progress indicator so you can see how many blocks have completed at a glance. (#6230)
- Selector and AI Fallback Fields in Action Blocks — Action blocks now expose explicit selector and AI fallback configuration, giving you finer control over element targeting. (#6212)
- Workflow Tagging API — Tags can now be created, applied, queried, and managed via the API for programmatic workflow organization. (#6251)
- Run Cost Visibility in Task Runs — Task runs now expose LLM, proxy, and captcha cost fields, making per-run spending trackable via the API. (#6219)
- Double-Click Action Support — The click action block now supports double-click for elements that require it. (#6166)
- Search on Credentials Page — Added a search bar to the credentials page for quick filtering by name.
Improvements
- “Inputs” Replaces “Parameters” Across the UI — Workflow block parameters are now consistently labeled “Inputs” throughout the editor and run views. (#6296)
- Sidebar Flyout on Hover — Hovering over a collapsed sidebar icon now reveals the flyout menu without requiring a click. (#6270)
- Copilot Follows Live Frontier During Runs — The copilot chat panel now tracks the active workflow frontier as the run progresses, keeping you oriented on the agent’s current work. (#6283)
- Completed Copilot Turns Collapse to Summary Card — Finished copilot turns now collapse into a compact summary card, reducing clutter in long sessions. (#6280)
- Copilot Workflow Proposal Staged View — Accepting or rejecting a copilot workflow proposal now shows a staged diff view that snaps back cleanly on cancel or reject. (#6260)
- Browser Profile Saving Indicator — A placeholder row now appears while a browser profile is being saved, giving immediate visual confirmation that the save is in progress. (#6215)
- Copilot Loop Detection — The copilot now detects when it is stuck repeating the same reply and actively rewrites its approach to break the loop. (#6161)
Bug fixes
- Fixed credentials with embedded multi-placeholder secret references not being resolved correctly during workflow runs. (#6245)
New features
- Guided Browser Profile Setup — A new step-by-step flow walks you through creating a saved browser profile, making it easier to set up persistent login sessions. (#6043)
- MCP Setup in Cloud Settings — Added a dedicated MCP setup section to cloud account settings for easier MCP client configuration.
- Draggable Workflow Blocks — Blocks in the workflow editor can now be reordered by dragging and collapsed to save space, making it easier to navigate complex workflows.
- CDP Connect Headers — The workflow editor now exposes a settings field for custom CDP connect headers, enabling authentication with remotely managed browser instances. (#6125)
- Credits Consumed per Run — Workflow run detail pages now show how many credits were consumed by that specific run. (#6133)
- Bulk Cancel Runs — A new API endpoint lets you cancel multiple in-progress workflow runs in a single request.
Improvements
- Archived Artifact Status — Workflow runs now show when their artifacts have been archived, with a clear status indicator in the UI. (#6075)
- Automatic Webhook Retry on Infrastructure Failures — Workflow webhooks are now retried automatically after transient infrastructure failures, reducing missed notifications. (#6062)
- Faster Script-Mode Runs — Speculative extract-action steps are skipped in script-mode workflows, cutting unnecessary LLM calls. (#6057)
- LLM Router Length-Truncation Fallback — When a model response is cut off by a length limit, the router now automatically retries with a fallback model instead of returning a partial response. (#6079)
- Analytics Range Extended to 365 Days — The analytics dashboard now offers a “365 days” time range, replacing “All time” which was slow for high-volume accounts.
- Copilot Recovers from Unexpected Errors — The workflow copilot now catches unexpected errors mid-run and presents a recovery option instead of silently stopping. (#6143)
- Copilot Surfaces Missing-Context Fallback — When the copilot lacks enough context to proceed, it now shows a clear fallback message explaining what information is needed. (#6144)
- Reduced LLM Context Overflow Failures — Prompt data (element trees, extraction goals) is now capped before being sent to the LLM, significantly reducing context-overflow and rate-limit errors on complex pages. (#6098)
Bug fixes
- Fixed popup pages opened during a run not being included in session recordings. (#6078)
- Fixed XHR-triggered file downloads not being captured when using download action blocks. (#6054)
- Fixed transient download-error popups incorrectly causing run failures. (#6080)
- Fixed child tasks not being terminated and workflow runs getting resurrected after a cleanup cron timeout. (#6047)
- Fixed copilot feasibility check stalling when given diagnostic input for an existing workflow. (#6061)
- Fixed reCAPTCHA v2 invisible variant being misidentified as v3, causing solve failures.
- Fixed downloaded file links on workflow runs showing internal storage paths instead of actual filenames. (#6137)
- Fixed the workflow copilot stopping unexpectedly when a browser session becomes unresponsive. (#6106)
New features
- Analytics Dashboard — A new /analytics page provides a breakdown of workflow run activity and performance over time. (#5980)
- Browser Profile Management Page — A dedicated browser profiles page lets you view, search, and rename your saved browser profiles. (#5990)
- Smarter Element Targeting — The agent now uses deterministic HTML element-tree compression, improving selector reliability on complex pages. (#5970)
- Block-Scoped Parameter Prompts in Debugger — The workflow debugger now prompts for parameters scoped to the specific block being tested, making per-block runs easier to set up.
Improvements
- Scheduled Trigger Badge on Runs Page — The global Runs page now shows a badge on runs triggered by a schedule, making scheduled runs easy to identify. (#5948)
- Signed Artifact Content URLs — Artifact content URLs are now always signed with configurable expiry for secure access control. (#5940)
- Faster Workflow Run API — The GET workflow run endpoint has been significantly optimized for speed. (#5993)
- Copilot Diff View Detects Nested Changes — The Copilot diff panel now correctly highlights changes inside nested conditional and loop blocks. (#5945)
- Output Parameter Size Protection — Workflow output parameter values are now capped to prevent oversized JSON payloads from causing downstream failures. (#5974)
Bug fixes
- Fixed scheduled workflow runs not appearing on the global Runs page. (#5949)
- Fixed webhook delivery silently failing on transient 403, 5xx, or network errors. (#5952)
- Fixed artifact downloads failing for clients that send HTTP Range request headers. (#5982)
- Fixed FileParserBlock failures not appearing in workflow run outputs. (#5964)
While Loop block, multiple schedules per workflow, Google Sheets connector, browser session runs tab
New features
- While Loop Block — A new workflow block type that repeats a sub-workflow until a condition is met, enabling dynamic iteration without a fixed loop count. (#5809)
- Multiple Schedules per Workflow — Workflows can now have multiple independent schedules (e.g., run at 9am on weekdays and noon on weekends). (#5799)
- Google Sheets Connector (Open Source) — The Google Sheets connector authentication flow is now part of the open source package. (#5815)
- Runs Tab on Browser Session Page — The browser session detail page now includes a Runs tab listing all workflow runs associated with that session. (#5830)
- Per-Workflow Total Step Cap — A new workflow-run step cap lets you limit the maximum number of agent steps across all blocks in a single run. (#5834)
Improvements
- TOTP Scoped to Active Credential — Credential-derived TOTP values are now scoped to the active credential, preventing stale tokens from being used mid-run. (#5805)
- Browser Dialogs Surfaced to Agent — JavaScript dialogs (alerts, confirms, prompts) that the browser blocks are now surfaced to the agent so it can adapt its next action. (#5810)
Bug fixes
- Fixed the workflow-deleted banner not appearing on the run detail page when the parent workflow has been deleted. (#5800)
MCP CLI for schedules and config, copilot hard-cancel, cursor in recordings, artifact URL improvements
New features
- Workflow Schedule Management via MCP — Manage workflow schedules from MCP clients (e.g., Claude Code) using the new
skyvern scheduleCLI commands. (#5713) - Org Settings via MCP — Configure organization settings from MCP clients using the new
skyvern configCLI tool. (#5694) - MCP OAuth Org Selection — Multi-org users can now choose which organization to connect to during MCP OAuth authentication. (#5699)
- Platform-Level JS Dialog Handler — The agent now handles JavaScript dialogs at the browser level using LLM-guided decision logic, improving reliability on dialog-heavy pages. (#5679)
- Print-to-PDF in Downloaded Files — Workflow blocks that trigger a browser print-to-PDF now expose the resulting file in the block’s downloaded files list. (#5784)
Improvements
- Cursor Visualization in Run Recordings — Playwright recordings now overlay a visible cursor, making it easy to follow the agent’s actions when reviewing a recording. (#5683)
- Short Signed Recording URLs — Browser session recording links are now compact signed URLs instead of full S3 paths. (#5686)
- Configurable Artifact URL Expiry — A settings field lets you control how long artifact content URLs stay valid before expiring. (#5690)
- Copilot Hard-Cancel — A hard-cancel button immediately stops an in-progress copilot run without waiting for the current step to finish. (#5726)
- Copilot Reuses Live Debug Browser Session — When testing a workflow from the debugger, the copilot now reuses the open browser session instead of starting a new one. (#5709)
- Copilot Surfaces WIP Workflow on Timeout — If the copilot times out or a test fails, it now offers the in-progress workflow so you can review and continue. (#5755)
- Pre-Filled Debugger Run Dialog — The Run dialog in the workflow debugger now pre-fills with values from your last run. (#5724)
- Copilot Atomic Workflow Accept — Accepting a copilot workflow proposal now applies all block changes atomically, preventing partial updates. (#5695)
- Loop Block Failure-Handling Clarity — The loop block UI has clearer controls and labels for configuring how failures inside the loop are handled. (#5731)
Bug fixes
- Fixed the in-progress workflow not being preserved when closing the copilot mid-run. (#5787)
Improvements
- Copilot Live Status Narration — The workflow editor copilot now displays a short live status line while it works, so you can see what the agent is doing instead of waiting in silence. (#5567)
-
Renamed Browser Profile Reset Action — The API endpoint for clearing a persisted browser session has been renamed from
/browser_session/refreshto/browser_session/reset_profilefor clarity. The old path remains available as a deprecated alias. (#5572) - Fewer Screenshots for API Runs — Screenshot capture frequency has been reduced for API-triggered runs, lowering storage overhead on long-running automations without affecting interactive sessions. (#5565)
Bug fixes
- Fixed the workflow copilot incorrectly proposing a workflow it had just failed to build. The copilot now only surfaces an Accept/Reject panel when the last test actually passed. (#5574)
- Fixed 2FA entry failing on pages with custom overlay input components where the real input field is hidden. The agent now types digit keys directly when no interactable input exists. (#5568)
Saudi Arabia proxy, workflow error mapping, smarter caching, MCP OAuth, AI output summaries, and reliability fixes
New features
-
Saudi Arabia Proxy Location — Added
RESIDENTIAL_SAas a proxy geolocation option for automations targeting Saudi Arabian websites. -
Workflow-Level Error Code Mapping — Workflows now support an
error_code_mappingfield that all blocks inherit automatically. Block-level mappings still take precedence, giving you a single place to define default error handling without repeating it on every block. (#5506) - AI Summarize for Workflow Run Outputs — A magic-wand “Summarize with AI” button now appears next to block and workflow run output sections, generating a plain-language summary of raw JSON output on demand. (#5556)
-
Reset Browser Profile API — A new
POST /workflows/{workflow_permanent_id}/browser_session/reset_profileendpoint clears a corrupted persisted browser session so the next run starts from a fresh browser state. (#5544) - MCP OAuth for Remote Clients — MCP clients (e.g. Claude Code, Codex) can now complete browser-based OAuth to authenticate with Skyvern without getting stuck in redirect chains. (#5558)
-
OpenClaw MCP Setup —
skyvern setup openclawandskyvern mcp switchnow support OpenClaw as an MCP client target for both remote and local stdio configurations. (#5536)
Improvements
- Smarter Extraction Cache Keys — The extraction cache now normalizes session-volatile tokens (random UUIDs, CSRF values, nonce query params, timestamps) before hashing cache keys. Semantically identical page visits now share cached extraction results, improving cache hit rates across workflow re-runs. (#5504)
-
Reduced Rate-Limit Errors on Extraction Prompts — Extraction prompts are now capped in size before being sent to the LLM. Oversized
previous_extracted_information,extracted_information_schema, and element trees are truncated rather than sent whole, significantly reducing429 RESOURCE_EXHAUSTEDerrors on high-volume workflows. (#5502) - Cross-Run Extraction Cache — Extraction results are now cached in Redis across workflow runs, scoped to the workflow. Recurring scheduled workflows skip redundant LLM calls when the page content hasn’t changed. (#5562)
- Copilot Survives Tab Switches and Reconnects — The workflow copilot agent now runs to completion even if the browser tab is closed or the connection drops mid-request, and the AI reply is persisted so it appears when you reconnect. (#5560)
- Copilot Fails Fast on Unresolvable URLs — The workflow copilot now immediately surfaces the real error (e.g. DNS resolution failure) and stops retrying when a URL cannot be reached, rather than reporting success. (#5563)
Bug fixes
- Fixed CSV parsing failing on files whose first header row exceeded ~1 MB, causing valid wide-format CSVs to be rejected with “Could not determine delimiter”. (#5510)
- Fixed a browser automation crash on sites with cross-origin stylesheets (e.g. MUI-based apps). CSS hover styles are now parsed entirely in memory without modifying the DOM, preventing React reconciliation errors that caused the agent to loop until max steps. (#5508)
- Fixed context-window overflow failures being misdiagnosed. Steps that fail due to oversized input are now correctly counted and short-circuit to a clear
LLM_ERRORfailure reason instead of triggering a redundant LLM-based failure analysis. (#5507) - Fixed workflow file-download blocks losing their configured
download_suffixat task finalization, causing downloaded files to be persisted without the expected filename suffix. (#5499) - Fixed cached scripts for non-ATS workflows incorrectly generating
fill_form()calls, which defeated the performance benefits of caching by delegating back to AI at runtime. (#5497) - Fixed the pagination navigator on the Workflow Past Runs and Run History pages being right-aligned instead of centered.
- Fixed run recordings missing an end timestamp and broken scrubbing. Recordings are now finalized with a stream-copy remux so players correctly show duration and support seeking. (#5551)
- Fixed tag input fields (e.g. skill set fields on job application forms) not selecting values from the dropdown. These fields are now correctly treated as auto-completion inputs. (#5557)
- Fixed per-step video sync spawning a redundant ffmpeg process on every step, causing CPU spikes and log noise. (#5559)
- Fixed workflow parameter edit and delete buttons being hidden when a parameter name was too long.
- Fixed webhook delivery failing silently when a stored webhook URL had a leading or trailing space. (#5550)
New features
- Adaptive Caching (Script Generation) — Skyvern can now learn from repeated workflow runs and generate cached scripts that speed up future executions. When a block runs successfully, Skyvern records a reusable script and replays it on subsequent runs, falling back to the AI agent only if the script fails. (#4908, #4916, #4917, #4920, #4922, #4931)
- Workflow Trigger Block — A new block type that lets one workflow trigger another from within the workflow editor. Chain workflows together as building blocks for complex automations. (#4885)
- Browser-based CLI Signup — New users can sign up for Skyvern Cloud directly from the CLI. Running
skyvern signupopens a browser flow that creates your account and stores your API key locally. (#4925) - Interactive ngrok Tunnel —
skyvern browser servecan now create an ngrok tunnel automatically, making it easy to expose your local browser to Skyvern Cloud without manual network configuration. (#4924) - South Korea Proxy Location — Added
RESIDENTIAL_KRas a proxy geolocation option for automations targeting South Korean websites. (#4918) - Downloaded Files Tab — Browser session detail pages now include a Downloaded Files tab for viewing and accessing files captured during a session. (#4911)
- CDP Proxy Authentication — Remote browser connections now support authenticated proxies via the CDP
Fetch.authRequiredprotocol. (#4936) - Remote Browser Download Interception — File downloads triggered by XHR requests and other non-navigation events are now captured when using remote CDP browser connections. (#4906, #4921, #4934)
current_dateReserved Parameter — Workflows now have a built-incurrent_dateparameter that resolves to today’s date automatically. (#4854)- Reserved Parameters in Block Editor — The workflow block parameter picker now shows reserved system parameters, making them discoverable without memorizing names. (#4857)
- Natural Language Loop
data_schema— Loop blocks driven by natural language extraction now support adata_schemafield for consistent output structure. (#4851) - Gemini 3.1 Pro and Inception Mercury-2 — Two new LLM engine options for your automations. (#4847)
- MCP Server on API Service — The MCP remote server is now mounted at
/mcpon the existing API, eliminating the need for a separate process. (#4843)
Improvements
- Renamed “Login-Free” to “Saved Profile” — Clearer terminology throughout the UI. Browser session checkbox wording has also been improved. (#4914)
- TOTP Verification Improvements — Better timer handling, expired code retry, 2FA banner suppression, and more reliable goal-text extraction during TOTP flows. (#4860)
- TOTP Webhook includes
workflow_permanent_id— Easier to identify which workflow is requesting a 2FA code. (#4871) - Consolidated Gemini 3 Pro Model Key —
VERTEX_GEMINI_3_PROnow auto-resolves to the latest version, so you no longer need to update config when Google releases point versions. (#4926) - Browser Rotation for Remote Connections — Remote browser environments now support context rotation, improving reliability for long-running automations. (#4929)
- Workflow save validation (HTTP 422) now returns clear, field-level error messages. (#4852)
- Browser launch errors now show actionable messages instead of raw exceptions. (#4844)
- Cloud LLM fallback overrides are now properly applied when selecting models. (#4839)
Bug fixes
- Fixed a race condition where final task status could be overwritten by a late-arriving non-final status. (#4928)
- Conditional blocks no longer crash when evaluating empty or null parameters. (#4907)
- Browser sessions using
FORCE_BROWSER_SESSIONno longer time out prematurely. (#4903) - SVG-based captcha text is now correctly extracted during DOM scraping. (#4880)
- MCP remote tool handlers now route to the correct API environment. (#4869)
- MCP remote auth token comparison no longer fails on encrypted tokens. (#4863)
skyvern quickstartnow handles local PostgreSQL without a pre-existingskyvernrole. (#4878)- WebSocket URLs through ngrok tunnels now rewrite correctly for live browser streaming. (#4927)
- Conditional blocks no longer crash when all branches use Jinja templates. (#4849)
- Conditional blocks no longer misinterpret resolved Jinja template variables. (#4841)
Unified Browser Task block, browser profile testing, 2FA detection, Skills package, Claude Opus 4.6, MCP block discovery
New features
- Browser Profile Testing UI (Frontend) — The Browsers page now lets you launch a test run against a saved browser profile to confirm its login state is still valid before scheduling production workflows. (#4833)
- 2FA Verification Code UI and Notifications — When a workflow pauses for a 2FA code, the UI now shows a clear prompt and pushes a notification so you don’t miss it. (#4787)
- Gated Admin Impersonation for MCP API Keys — Adds gated admin controls so support engineers can impersonate an organization scope when troubleshooting MCP-based automations, with audit-friendly access controls. (#4822)
- Browser Profile Testing — Test browser profiles directly from the UI to verify that saved login sessions are still valid. Workflows can also run with a saved browser profile, enabling login-free automations that reuse authenticated sessions. (#4818)
- Skyvern Skills Package — The new
skyvern skillCLI command provides pre-built workflow templates and reference documentation for common automation patterns. (#4817) - Clear Cached Scripts API — A new endpoint lets you clear cached automation scripts for a workflow when a target website changes and you want to force Skyvern to re-learn. (#4809)
- Automatic 2FA Detection Without TOTP Credentials — Skyvern now detects when a website asks for a 2FA code even without pre-configured TOTP credentials. The system pauses and waits for you to provide the code via the UI or webhook. (#4786)
- Full CLI Parity with MCP — The CLI now supports browser session management, credential management, block operations, and enhanced workflow commands. Everything available through MCP is now accessible from the command line. (#4789, #4792, #4793)
- Claude Opus 4.6 CUA Support — Anthropic’s Claude Opus 4.6 is now available as a Computer Use Agent model for driving browser interactions. (#4780)
- Anthropic Claude Opus 4.6 — Added Claude Opus 4.6 as an available LLM engine for web automation tasks. (#4777, #4778)
- Unified Browser Task Block — The Navigation V1 and Task V2 blocks have been merged into a single Browser Task block. The block now exposes both engines as a dropdown, simplifying the workflow editor and reducing block-type confusion. (#4695)
- Block Discovery MCP Tools — Added MCP tools that let AI assistants enumerate available block types and their schemas, plus richer instructions for building workflows through MCP. (#4683)
- Edit Existing Credentials API — A new endpoint lets you update credential values without deleting and recreating them, preserving credential IDs across changes. (#4693)
- Filter Workflow Runs by Error Code and Search Key — The workflow runs list endpoint now supports
error_codeandsearch_keyquery parameters for narrowing down failed runs. (#4694) - GPT-5 Mini in Model Selector — GPT-5 Mini is now selectable as an LLM engine. GPT-5 has been hidden in favor of more reliable variants. (#4686)
Improvements
- 2FA Backend Cleanup — Refactored the 2FA pipeline to make code delivery more reliable across both the UI form and the API endpoint. (#4826)
- Environment Variables Override Mounted API Keys — In Skyvern’s Streamlit-based deployment, env vars now take precedence over keys baked into the image, making local overrides simpler. (#4824)
- The workflow run timeline now properly displays loop iterations and conditional branches, making complex runs easier to debug. (#4782)
- Block Failure Reasons Surfaced in UI — Every block type now shows its
failure_reasonin the run timeline, making debugging non-obvious failures faster. (#4672) - Webhook Payload Truncation in Logs — Large webhook payloads are now truncated in log entries, keeping logs readable without losing context. (#4701)
Bug fixes
- When both a TOTP credential and a webhook are configured for 2FA, the credential is now correctly prioritized. (#4811)
- Fixed conditional blocks evaluating the wrong value after Jinja template rendering. (#4801)
- Fixed MFA resolution priority when both TOTP credentials and webhooks are configured. (#4800)
- Fixed the cursor jumping to the end while editing conditional expressions in the workflow editor. (#4682)
- Fixed dotted Jinja variables (e.g.
{{ block_output.field }}) rendering incorrectly in the expression preview. (#4684) - The API now returns 404 instead of 500 when a workflow is not found. (#4699)
New features
- Conditional Block Caching — The script cache now supports Conditional blocks, so workflows that branch based on page state can replay from cache instead of re-running the AI agent. (#4642)
- OpenTelemetry Backend — Skyvern can now export metrics, traces, and logs to any OTEL-compatible collector for centralized observability. (#4632)
- Workflow API Route Reorganization — Workflow API endpoints have been split into focused routers, making the OpenAPI surface easier to navigate. (#4628)
Improvements
- Workflow Copilot Auto-Opens for Empty Workflows — When you create a new workflow with no blocks yet, the Workflow Copilot opens by default to help you get started. (#4625)
- Run Workflow Button Moved to Top — The Run button now lives at the top of the workflow editor for easier access. (#4626)
- Skip LLM for Single-Match Location Autocomplete — Location autocomplete now skips the LLM call when only one option matches the user’s input, reducing cost and latency. (#4631)
- Improved Version History Icon — The workflow version history button now uses an icon that better represents document versions. (#4640)
Bug fixes
- Fixed
ForLoopblocks failing to execute under script caching (regression from v1.0.12). (#4630) - Fixed
ActionDragPathsubscript error in CUA drag-action parsing. (#4623) - Fixed cron-scheduled tasks getting stuck with “task not found” errors. (#4627)
- Fixed parameter creation dialog not clearing default-value state when switching to credential type. (#4622)
- Fixed debugger loop-block visual overlap when the status row appeared. (#4641)
New features
- Task V2 Termination — Task V2 (the planner-agent-validator system) can now be terminated mid-run via API, matching the cancel behavior already available for Task V1. (#4589)
-
ForLoop Block Caching — Cached scripts now support
ForLoopblocks. Workflows that iterate over lists can replay from cache after the first successful run. (#4600) - Conditional Block Caching — Initial support for caching workflows that include conditional branching. (#4603)
-
run_withParameter for Run Task — Therun_taskendpoint now accepts arun_withparameter so you can target Task V2 (or future variants) without changing the endpoint. (#4576) -
skyvern run devCommand — A new CLI command spins up a Superset workspace alongside the standard dev environment for local analytics work. (#4593)
Improvements
- All Branch Evaluations Shown in Conditional Block Timeline — The run timeline now shows every conditional branch’s evaluation result, not just the chosen one — useful when debugging “why did it pick that branch?” (#4577)
- Sidebar Expand/Collapse Icon — Replaced the pin icon with a chevron, matching common conventions. (#4586)
- PDF Parsing Token Cap — PDF parsing now caps token volume per request to avoid total failure on very large documents. (#4602)
-
DATABASE_STRINGParameter on Quickstart —skyvern quickstartnow accepts aDATABASE_STRINGargument for explicit DB connection configuration. (#4580)
Bug fixes
New features
- Workflow Copilot Generally Available — The Workflow Copilot is now enabled for all users. Describe what you want in natural language and it generates a multi-block workflow you can edit. (#4533)
- Browser Sessions v2 Frontend — The new browser session management UI lands on the Browsers page, building on the v2 backend shipped in v1.0.10. (#4514)
-
TOTP Endpoint OTP Type Filter — The TOTP listing endpoint now accepts an
otp_typequery parameter for filtering numeric codes vs. magic links. (#4529) - Mintlify Documentation Setup — Skyvern’s documentation site moves to Mintlify, with redirects for the legacy Fern URLs. (#4516)
Improvements
-
Async File I/O via aiofiles — Replaced blocking file I/O with
aiofilesthroughout the agent runtime, improving throughput on I/O-heavy workloads. (#4520) -
Cached Experimentation Provider TTL —
BaseExperimentationProvidernow caches feature-flag lookups with a per-item TTL, reducing pressure on the experimentation backend. (#4536, #4540)
Bug fixes
- Fixed
ai_fallbacknot being read correctly from the database for cached workflow runs. (#4524) - Fixed iframe-tree extraction crashing when the page raised a transient exception. (#4530)
- Fixed the workflow version comparison view ordering versions on the wrong sides. (#4531)
- Fixed the workflow history panel scroll behavior in debug mode. (#4532)
- Fixed
TypeErrorwhen comparing offset-naive and offset-aware datetimes during workflow scheduling. (#4534) - Fixed schema regeneration overwriting unchanged block field names. (#4535)
- Fixed
uvicornreload breaking on Windows due to async event loop policy. (#4538)
New features
- Browser Sessions v2 (Backend) — A new persistent browser session architecture with cleaner lifecycle management, better resource reuse, and improved reliability for long-running automations. The frontend follows in v1.0.11. (#4515)
- General Text CAPTCHA Solver — Skyvern can now solve a wider range of text-based CAPTCHAs without site-specific configuration. (#4517)
- TOTP Webhook Retries — TOTP webhook delivery now retries automatically on transient failures, reducing the rate of “no code received” errors during 2FA flows. (#4518)
- Block-Label Validation in
run_blocks— Therun_blocksendpoint now validates block labels up front, returning a clear error instead of failing partway through execution. (#4500) - Print PDF Block — A new workflow block that prints the current page to PDF and saves it to your run’s downloaded files, useful for archiving receipts, invoices, and reports. (#4452)
- HTTP File Downloads — The HTTP Request block can now download files directly via HTTP (not just GET JSON), enabling workflows that pull artifacts from authenticated APIs. (#4440)
- Workflow Copilot Streaming with Cancel — The Workflow Copilot now streams its responses with a cancel button, so you can stop generation mid-stream if you want to refine your prompt. (#4437, #4456)
- “Execute on Any Outcome” (Finally) Block Option — Blocks can now be configured to run regardless of whether prior blocks succeeded, enabling cleanup steps and “always send report” workflows. (#4443)
- Persistent Browser Session Uptime Billing — Billing v2 now meters persistent browser session uptime separately, with PostHog feature flags controlling rollout. (#4444)
- GPT-5 Mini Flex Engine — Added GPT-5 Mini Flex as an available LLM engine. (#4432)
- Clipboard Copy over HTTP — The browser now exposes clipboard copy functionality that works in HTTP-served browser sessions, not just HTTPS. (#4446)
Improvements
- Confirmation Dialog Shows Affected Blocks — Deleting a parameter or block now lists every block that references it, so you don’t accidentally break downstream logic. (#4519)
- Workflow Updates Return 409 on Race — Concurrent workflow updates now return HTTP 409 instead of silently overwriting each other. (#4510)
- Tiktoken Caching — Tokenizer instances are now cached, cutting per-step overhead on token counting. (#4521)
- HTTP Block Boolean and Integer Templates — The HTTP Request block now correctly passes boolean and integer template values without coercing them to strings. (#4435)
- Bitwarden and Azure Vault Tooltips — Added inline help tooltips explaining how to set up Bitwarden and Azure Key Vault credential sources. (#4447)
- Smaller S3 Objects Use STANDARD Tier — Storage cost optimization: small artifacts now use STANDARD instead of GLACIER_IR. (#4453)
- Reduced Max SVG Parsing — Lowered the SVG parsing cap to prevent runaway memory usage on pathological pages. (#4430)
Bug fixes
New features
- Workflow Copilot v1 — The first version of the Workflow Copilot ships behind a feature flag. Describe a workflow in natural language and the copilot generates blocks, parameters, and connections you can edit on the canvas. (#4401, #4413, #4416, #4423)
- Per-Action Credit Billing (Billing v2) — A new billing model meters credits per browser action instead of per step, giving more predictable cost behavior on action-heavy workflows. (#4398)
- Azure Key Vault Secret Storage — Skyvern can now store and retrieve credentials from Azure Key Vault as a managed secret backend. (#4384)
- Conditional Inside Conditional — Conditional blocks can now be nested inside other conditional blocks for richer branching logic. (#4362)
-
Reference Workflow Output in Sub-Blocks — Blocks can now reference upstream
workflow_outputvariables, simplifying multi-stage workflows. (#4414) - Browser Type Choice for Persistent Browser Sessions — Persistent sessions now let you choose the browser type (Chromium, etc.) at create time. (#4406)
- Browser Extension Support for PBS — Persistent browser sessions can launch with a specified browser extension. (#4364)
-
Step Count in Webhooks and Run Payloads — Webhook payloads and
get_runresponses now includestep_countfor cost reconciliation. (#4410) - PH (Philippines) Proxy Location — Added Philippines as a residential proxy option. (#4395)
- TextPromptBlock Artifact Storage — TextPromptBlock now saves the LLM prompt and response as run artifacts for debugging. (#4361)
- Action–Screenshot Linking — Actions in the run timeline now link directly to the screenshot captured at the time of execution, so you can see what the agent saw when it made each decision. (#4403, #4404)
Improvements
- Clearer Errors for Missing Workflow Parameters — Missing required parameters now return a clear HTTP 400 instead of HTTP 500. (#4380)
-
Empty-String Defaults for String Parameters — Workflow string parameters without explicit defaults now initialize to
""instead ofnull, matching common Jinja-template expectations. (#4382) -
HAR Files Use Zstd Compression — HAR archives now ship as
.har.zst, cutting storage size significantly with no fidelity loss. (#4420) -
Hydrated Workflow Title in Run Responses —
workflow_runresponses now include the workflow title for cleaner logging without an extra API call. (#4363) -
Cached Workflow-Script Lookup —
get_workflow_script_by_cache_key_valueis now cached and post-action execution can re-enable code generation. (#4385) - Updated Datacenter Proxy Provider — Datacenter proxies now route through a new provider with broader IP coverage. (#4386)
- Improved Conditional NL Branch Tooltip — Updated tooltip and placeholder text on conditional blocks to better describe how natural-language branches are evaluated. (#4368, #4369)
-
OLLAMA Vision Support Toggle — Added
OLLAMA_SUPPORTS_VISIONenv var to enable vision-capable Ollama models. (#4351)
Bug fixes
- Fixed older credentials becoming inaccessible after an internal schema change. (#4358)
- Fixed scrolling actions misfiring on pages with overflow-hidden ancestors. (#4389)
- Fixed
close_browser_on_completionnot being honored during workflow run cleanup. (#4381) - Fixed run-ID generation collisions under high concurrency. (#4383)
- Fixed empty browser-extension config crashing PBS launch. (#4365)
- Fixed wrong screenshots being attached to certain actions. (#4400)
- Fixed nested DB connection management in workflow execution. (#4402, #4411)
- Fixed
local variable not definederrors in cached-action execution. (#4376) - Fixed artifacts endpoint incorrectly including
api/in the path. (#4417) - Fixed S3 storage class selection for small objects (
_get_storage_class_for_org). (#4424) - Fixed cached actions executing without a brief settling delay. (#4418)

