Skip to content

November 2026

Organization notifications can open alerts in a PagerDuty service. Each alert’s PagerDuty severity follows the risk’s urgency or the incident’s severity, an import of many records pages once instead of once per record, and alerts never include descriptions or comment text. Adversarial opens alerts and never resolves them. Set up PagerDuty.

Import risks from Cloudflare Security Center insights, Zero Trust CASB posture findings, and Page Shield malicious scripts, and incidents from WAF attack campaigns and Brand Protection lookalike domain and logo matches. A single connection, with your Cloudflare account ID and an API token, can serve both, or you can use separate connections whose API tokens have different permissions. Each integration chooses which sources it imports. Each source syncs on its own, so a source that Cloudflare refuses does not stop the others. A WAF incident is SEV-4, or SEV-5 when it may be a real user, and its title and status say whether some requests succeeded. Set up Cloudflare.

Import Semgrep findings as risks: Code (SAST), Supply Chain (SCA), and AI-powered detection findings, each as its own source you can switch on or off, with filters for issue type and severity. Semgrep triage states carry over: open and reviewing findings are New, fixing is Remediation, provisionally ignored is Closure Proposed, and ignored or fixed is Closed. A finding that disappears from Semgrep closes its risk. Set up Semgrep.

Import Sublime Security flagged message groups as incidents with the source “Email Protection”. Groups that Sublime reviewed automatically are skipped, and groups a person reviewed import as Closed, except a Skip review, which is In Progress. Attacks import without a severity, so AI scoring or a person rates them; a group that is not an attack, such as one reviewed as benign, creates no incident, and an incident already imported for it closes at SEV-5. An unreviewed attack that nobody opened or reported closes, and reopens if a recipient opens it. Each description lists the flagged rules’ attack types and tactics, any employee or executive impersonation, and what recipients did with the email. Only Sublime Cloud accounts are supported. Sublime Security.

GitHub Code Scanning and Secret Scanning Alerts

Section titled “GitHub Code Scanning and Secret Scanning Alerts”

The GitHub integration imports code scanning and secret scanning alerts as risks, alongside Dependabot alerts. Each GitHub integration chooses which alert kinds it imports, and a kind your token can’t read is skipped while the others still import. Code scanning imports security alerts only, and secret scanning never reads or stores secret values. Existing integrations keep importing Dependabot alerts. GitHub.

When a risk or incident integration’s latest sync didn’t succeed, a line under its name on Settings > Integrations says why. Sync Error, in red, means the source refused the sync, for example because the credentials are missing a permission, and syncing stops until someone fixes it. Retrying means a passing problem, such as a rate limit or an outage at the source, and the next sync tries again. The reason includes the source’s own explanation where it gives one, such as the scopes a Wiz service account is missing.

When an integration first stops with a Sync Error, members who can manage integrations get an in-app notification, “Integration name stopped syncing”, with the reason. It is sent once until a sync succeeds, and never by email, Slack, Teams, or PagerDuty. Sync errors.

The Risk Assessment Management Procedure (RAMP) scope is now a checklist instead of a Cyber or Cyber + Operational Risk switch. Cyber is always on, Operational Risk works as before, and a new Third-party Risk Management option adds a Sponsor role, third-party onboarding guidance, and Third-party Control Assurance scoring guidance for risks whose Source is TPRM. Organizations that chose Cyber + Operational Risk keep Operational Risk ticked. The default risk source previously named “Vendor Management” is now TPRM, and risks and saved views that used the old name follow the rename. This gives teams a documented way to codify how third-party risk fits into their overall risk management process today. It is also the foundation for a full TPRM module, coming soon, that builds on these RAMP updates to automate the entire third-party risk management process and integrate it natively into the Risk Register. Scope and Third-party Risk Management.

A new Agent Context integration type lets AI scoring look things up in your connected tools while it scores a risk or incident: Jira, Linear, and ServiceNow tickets, GitHub code, and Wiz assets and identities. For example, scoring can check whether a risk already has a ticket, or whether flagged code appears in your repositories. Jira and Linear lookups can be limited to one project or team, ServiceNow lookups search the one table you choose, and on each integration you choose which lookups are allowed and use Test to try them. Agent Context is experimental and hidden by default: turn on Agent Context Integrations under Experimental Features in your Preferences to see it on your own screens. Manage your integrations.

On an imported risk or incident, click the locked Title or Description and select Stop Syncing to keep your changes. Syncs then ignore the record until you select Resume Syncing. The record shows who paused it and when. Stop syncing a record.

Turn on Sync comments on a Wiz or GreyMatter integration to import Wiz issue notes and GreyMatter incident comments as comments on their records. Each comment imports once, and its text names the person who wrote it in the source. Comments.

An open risk or incident has an Associations card below Remediation or Response that gathers everything linked to it: an Imported From row naming the integration item it came from, the linked records of the other type, its child records, and Attachments. On a risk, Related Incidents lists linked incidents and Related Risks lists its child risks; on an incident, Related Risks lists linked risks and Related Incidents lists its child incidents. The card’s heading shows how many associations there are. The Imported From row shows whether the record is still syncing with the source product or its sync is paused. Tickets stay in the Remediation card.

Each list you can edit has a + button. When it has only one action, such as linking an incident to a risk, + opens the search directly; on an incident’s Related Risks, it offers Link existing and Create risk. A linked record you don’t have access to shows a lock and “No access”, and a list for a record type you can’t read has no + button. On wide screens the left and right sides of the modal scroll separately. See incident related risks, linking risks, and linking incidents.

Tickets and Linked Records from AI Assistants

Section titled “Tickets and Linked Records from AI Assistants”

AI assistants connected through MCP can list, search, create, link, and unlink Jira, Linear, and ServiceNow tickets on risks and incidents, using your organization’s ticket integrations. They can also create a risk from an incident and link or unlink risks and incidents in either direction, and the risks and incidents they read now include their links. Assistants connected before this release must reconnect to use the ticket tools. MCP tools and workflows.

The nightly full sync looks up open records whose source items it did not see. GreyMatter, watchTowr, Bugcrowd, and HackerOne records whose source item is gone close with a comment. Cloudflare risks and incidents close when the insight, CASB finding, Page Shield script, or Brand Protection match is gone, including dismissed insights. A CrowdStrike incident closes when all of its Falcon alerts are closed or hidden, and GitHub code scanning and secret scanning risks close when their alert is fixed, dismissed, or resolved. Items removed from the source.


The Tickets section of a risk or incident has a + button in its header with Link existing and Create new. Create new opens a short form where you choose the Jira project, Linear team, or ServiceNow table and create the ticket, without searching first. The Related Risks section of an incident works the same way, with Link existing and Create risk. Viewers without edit permission can’t create tickets. See Jira, Linear, and ServiceNow.

Risks no longer have a Control Statement field. It is gone from the editor, the register’s columns and sorting, register search, CSV exports, History, and change notifications, and existing control statements have been removed. CSV files that still have a Control Statement column import, and the column is ignored. Risk fields.

CrowdStrike integrations no longer have a Sync setting. CrowdStrike now syncs like every other integration: every 15 minutes for new and updated alerts, plus a full sync each night. CrowdStrike syncs also no longer fail when Falcon returns an alert whose user details are incomplete. CrowdStrike.

Editing a cell in the Risk or Incident Register, such as Source, Status, Severity, tags, or the assignee, shows about ten options at a time instead of five.

The Portfolio View loads its incident severity counts faster for organizations with many incidents.

Settings > Company is now Settings > Organization, and its Company Name field is Organization Name. Bookmarks to the old Company settings address no longer open the page; open it from the Settings menu instead.

A role that already includes another role’s permissions, such as Editor including Viewer, is assigned on its own. On Settings > Team, included roles show as covered and can’t be selected together with the role that includes them, and existing members, invitations, and service accounts drop the redundant roles. Role summaries read, for example, “Includes permissions of Viewer”. Nobody loses access. Assigning roles.

Sublime Security, CrowdStrike, and Cloudflare WAF incidents that a vendor control stopped now get a Contained Date, so reports count them as contained. A Sublime message group is contained from when it was quarantined, trashed, or moved to spam, and the date clears if Sublime restores it to the inbox. A CrowdStrike incident is contained when every alert’s process or file was killed, blocked, or quarantined, and a fully blocked Cloudflare WAF campaign is contained when detected. A Contained Date that a person set always stays. See Sublime Security, CrowdStrike, and Cloudflare.

Wiz issue notes are no longer added to risk descriptions. They import as comments when Sync comments is on, and are not imported when it is off. Risks that showed notes in their description lose that section on their next sync. Wiz.

Changing an Integration’s Settings Runs a Full Sync

Section titled “Changing an Integration’s Settings Runs a Full Sync”

Changing what a risk or incident integration imports, such as its severities, states, or GitHub alert kinds, makes its next sync a full sync. Records that the old settings left out import on that sync instead of only once they next change in the source.

The pickers that link a risk or incident to another search titles and IDs, such as RSK-42, instead of every description and comment, and start after two characters, so results come back faster. Register search on a broad term in a large register returns sooner, and a multi-word search ranks titles containing every word, such as “Ransomware Outbreak” for outbreak ransomware, above description matches.

Scoring several risks or incidents with AI no longer asks you to review and save the results. The scores are saved as you, in a background job, and the register refreshes with a notification when it finishes. Single-item AI scoring still suggests a score for you to accept. See risk bulk scoring and incident bulk scoring.

AI scoring now takes into account a risk’s or incident’s tags and who opened it, and an incident’s source. A RAMP can now give scoring rules for tagged items, such as risks tagged OPS, and AI scoring applies them. Changing an item’s tags changes what its next AI score is based on.

Cloudflare CASB Instances and Ignored Findings

Section titled “Cloudflare CASB Instances and Ignored Findings”

Each Cloudflare CASB risk’s description lists its active instances with the asset, its type, and the date it was detected, and the link opens the finding’s instances in Cloudflare. The Discovered Date is the earliest detection date among the finding’s instances, including hidden ones, instead of the date of import, and risks already imported take that date on their next sync. A finding you ignore in Cloudflare now closes its risk; previously it stayed open. Cloudflare.

When you don’t have permission to edit, a risk or incident opens read-only: fields keep their full-contrast text without edit controls, every tag is shown, and actions you can’t use, such as Delete, Score, and the comment box, appear disabled with a tooltip. A “Read-only — contact your admin for edit access” line replaces the banner that was previously hidden. Before this change, edits and comments could look saved but weren’t.

Removed Source Items Close Without a Limit

Section titled “Removed Source Items Close Without a Limit”

The nightly sync no longer limits how many records it closes at once. Every open GreyMatter, watchTowr, Bugcrowd, or HackerOne record whose source confirms the item is gone now closes, even if that is all of them. Items removed from the source.

GreyMatter and Source Lookups Hold Up Under Limits

Section titled “GreyMatter and Source Lookups Hold Up Under Limits”

GreyMatter imports read comments only when Sync comments is on, and the nightly lookup of incidents a sync did not see never imports comments. When GreyMatter credits run out, the sync stops instead of failing and runs again after the credits reset. When the nightly sync looks up records whose source items it did not see, a failed lookup leaves only its own record open, to be looked up again next time, instead of its whole batch, for GreyMatter, watchTowr, Bugcrowd, and HackerOne.

Adding or removing a tag on a risk or incident, from the editor, the register, or a multi-edit, updates its last-updated date and appears in its History as “Added tag” or “Removed tag” with the tag’s name. Tag changes don’t send change notifications. See risk tags and incident tags.

The risk field labelled Class in the register and the editor is also called Class in CSV exports, the CSV import template, History, change notifications, filter view settings, and integration previews. CSV import accepts both a class column and the older type column, so earlier exports and templates still import. Risk classes.

The Cmd/Ctrl+K command palette matches risk and incident titles, ranks exact and whole-word title matches first, and starts searching after two characters, so results come back faster in large registers. It no longer matches descriptions, comments, or reordered words such as “cred vpn” for “VPN credential leak”; the register search still does. Typing a record ID such as RSK-42 still jumps straight to it, and the keyboard keeps working after you click inside the palette. Keyboard shortcuts and the command palette.

A sync no longer moves a risk in Urgency Proposed, Remediation, or Closure Proposed back to New when the source still reports the finding as open. A risk closed in Adversarial still reopens if the source reports it open again. Risk integrations.

Integration Syncs Keep a Person’s Incident Status

Section titled “Integration Syncs Keep a Person’s Incident Status”

An incident status that a person set last stays when a sync runs. Any other status follows the source, so an incident that the sync closed reopens when the source opens it again, and the Contained Date the sync set is cleared. Syncing records.

The GreyMatter States setting lists all nine GreyMatter states, and all are ticked by default. On hold, Pending rq, and Pending vendor map to In Progress, and Cancelled maps to Closed. A severity that a person set stays when GreyMatter closes an incident with a benign close code. GreyMatter.

CrowdStrike syncs import only shown alerts. When every alert of an open incident is hidden in Falcon, the incident closes with a comment, and it reopens if an alert is shown again. A sync does not change a severity that a person set. CrowdStrike.

In generated board decks and CyberGov reports, executive summaries and threat card descriptions stay inside their boxes, shrinking the text and then ending with “…” when they run long. Flow appendix and chart headers are evenly spaced, and same-day risks and month-end labels appear on the charts. A severity band with 23 or 24 incidents may now continue onto one more slide. Generated reports.

  • CSV exports of risks and incidents prefix any cell that begins with =, +, -, or @ with an apostrophe, so spreadsheet programs never run it as a formula. Importing an export back keeps the apostrophe.
  • Deactivating a member, or a member leaving, deletes the invitations they sent.
  • A role change deletes any invitations the new roles could no longer issue.
  • Descriptions, comments, titles, and remediation tasks written by people are limited to 65,536 characters.
  • Markdown too dense to display safely shows as plain text instead of freezing the browser.

The rich-text editor and chart code now load only on the pages that use them, so the sign-in page and Home dashboard paint sooner.

Opening a risk or incident from a searched register keeps your search term, and the Next and Previous arrows step through the search results instead of the whole register. What register search matches.

Line breaks in imported or typed Markdown descriptions, such as one field per line in a Google Workspace alert, display on separate lines in the editor and survive saving, instead of running together.

A Cloudflare CASB risk’s Discovered Date no longer resets on every sync, and Security Center insight links open the right Cloudflare page. Risks already imported take the earliest detection date among their instances on their next sync. Cloudflare.

GreyMatter Imports New Incidents by Default

Section titled “GreyMatter Imports New Incidents by Default”

GreyMatter integrations without a saved States selection now also import incidents in the New state. Integrations that already saved a selection keep it.

Clearing a mixed Assigned To value when editing several items together now unassigns every selected item. Previously the clear was ignored. Command palette entries for Profile, Invites, Portfolio, and Preferences, which used to open a missing page, now work.


Bulk Scoring Endpoints Save and Run as Jobs (Breaking)

Section titled “Bulk Scoring Endpoints Save and Run as Jobs (Breaking)”

POST /v1/risks/scores and POST /v1/incidents/scores now save the scores. They return 202 with a job ID, to follow through the jobs endpoint, instead of returning scores in the response, and an empty list of IDs is refused. Before this change they returned scores and saved nothing. The single-item score endpoints keep their paths and still only suggest a score. Their operation IDs are now suggest_risk_score and suggest_incident_score, so generated clients rename those functions. Clients that read scores from the bulk response need to read the job instead. See Score risks and incidents with AI and the API reference.

Stricter OAuth Requests (Breaking for Nonconforming Clients)

Section titled “Stricter OAuth Requests (Breaking for Nonconforming Clients)”

Apps that connect through OAuth, including MCP clients, must now send a well-formed scope with no stray spaces or tabs and a PKCE code verifier of 43 to 128 standard characters, or the request fails with invalid_scope or invalid_grant. The MCP server accepts a token for its /mcp resource whatever the letter case of the scheme and host. Refreshing a service account’s token now applies its current roles, and refreshing a deactivated service account’s token is refused.

Creating or updating a risk or incident through the API, MCP, or CSV import refuses a title or remediation task over 65,536 characters with a validation error, as descriptions already were. Clearing the closed date of a Closed risk returns 422.

Finished background jobs, such as imports and bulk scoring, are kept for 30 days, and failed jobs for 90 days. After that they no longer appear in the jobs endpoint. A failed job that the platform abandoned because it stopped responding has settled_by set to sweep, and its message says whether to request it again or whether the next scheduled sync picks it up.

Risks returned by the API and MCP no longer include control_statement. Requests that still send control_statement in a risk are accepted and the field is ignored, but sorting a risk list by control_statement now returns an error that lists the accepted sort fields.

Creating a risk through the API sets its assignee from assigned_to, the same name risk updates and incident creation use. It was previously ignored without an error. The older assigned_to_id still works.


  • A Closed risk always keeps its closed date, so registers and date-based metrics agree.
  • In the compliance editor, saving and approving can no longer run at the same time and lose edits, and leaving a version with unsaved changes asks first.
  • Slack notifications with a long title or details are shortened to fit instead of failing to send.
  • Slack, Teams, and PagerDuty notifications no longer wait behind long-running work in your organization before sending.
  • The Add Domain and Invite dialogs and the integration connection editor always open as dialogs, instead of sometimes opening full screen or with the previous window’s contents.
  • Pages no longer show a second scrollbar or blank space below the app, and wide tables on Settings > Team, the Roles view, and Settings > Notifications scroll inside their frames.
  • A Sublime Security message group that Sublime reviewed automatically closes an incident imported before the review.
  • A Cloudflare connection whose API token expired or was revoked asks for new credentials instead of reporting a missing permission on every sync.
  • On a risk or incident, an open ticket or related-item menu now stays in front of the add buttons below it instead of being hidden behind them.
  • CrowdStrike alerts in progress in Falcon import as In Progress.
  • A Cloudflare Brand Protection sync keeps a severity you set on the incident.
  • Register search finds comments and edited descriptions that contain characters such as _, *, [, or &, for example host_01.
  • The settings navigation and register popovers, such as filters, the tag picker, and the user picker, show thin scrollbars again instead of the wide green default.
  • Microsoft Teams channels in Settings > Notifications no longer show a # before their names; only Slack channels do.
  • When following or unfollowing a risk or incident fails, an error is shown and the follow state stays as it was, instead of looking saved.
  • A Cloudflare CASB risk closes on the next full sync when its CASB integration has been deleted in Cloudflare.
  • Pressing Escape in an open search picker on a risk or incident closes only the picker, not the risk or incident.
  • Product names and search in the Add Instance dialog on Settings > Integrations stay readable at every screen width.