SECUREHIVE

What's new

We're constantly improving SecureHive. Here's what's new, what's coming, and how it helps you stay secure and compliant.

Highlights for you

Security Portfolio
Available

Map your security stack to capabilities and risk scenarios—see coverage, gaps, and overlap in one place, with CSV export and an audit trail for leadership and audit conversations.

Incident management
Available

Full incident lifecycle: create, triage, assign, and track security incidents. Coordinate response and keep stakeholders informed.

Maturity model & CIS framework
Available

Assess security maturity with the CIS Controls framework. License from marketplace, assess by domain, track levels and applicability. More frameworks (CMMC, NIST CSF, ISO 27001) coming soon.

Transparent updates
Available

We document every release so you know what’s changed and when. Plan upgrades and stay in control of your environment.

Reliable releases
Available

We follow semantic versioning so you can plan upgrades with confidence. No surprises—just clear, predictable updates.

Release history

Unreleased

Fixed

  • /about/changelog showed "Changelog file could not be loaded" on deployed environments. The page reads CHANGELOG.md at request time, but .dockerignore excluded the file from the build context and the standalone output did not trace it. .dockerignore now keeps CHANGELOG.md, and next.config.ts adds it to outputFileTracingIncludes for the route. Reached by clicking the version number in the sidebar.

Added

  • iKredo fold-in, phases 1 to 4, frontend (frontend#659, #664, #665, #666; backend counterparts #966, #972, #973, #974). All screens are hidden until the matching switch is turned on for the workspace.
    • Reviews under Goals: cycles, reviews table, workbench, my review, feedback to give, policy.
    • Personal portal at /portal: sign-in by emailed code, wallet, credential offers and export, vendor console (/portal/vendors), security jobs (/portal/jobs, public model card at /portal/jobs/model-card).
    • Vendor windows under Vendor Risk, and Security jobs under Objectives & Outcomes (/s/[tenant]/goals/jobs).
    • Lists are EditableTable with TablePager and server-side search and paging; strings in en, es and ja. The login release-notes dialog announces 1.20.0.
  • Release 1.20.0 note for Control Observations. The login release-notes dialog (src/lib/release-notes.ts) gains a 1.20.0 entry for the Observe step and the screens that show a read (builder, asset "Controls on this asset", control test Evidence reads, the "Read from SIEM" badge, the approval card), and CURRENT_RELEASE_VERSION and package.json move to 1.20.0. The entry covers this feature only; the other 1.20.0 content merges into the same entry. Release notes are English-only data, so no locale file changes.
  • vCISO partner channel, frontend (phases 1 to 3; backend feat/partner-channel-phase1).
    • Partner console at /p, /p/[partner] (Clients), /p/[partner]/clients/new (Onboard a client), /team, /statements and /templates, in the route group app/(partner-console) outside the tenant shell with its own layout and bar (the firm's colour and logo when it has valid ones; branding is treated as untrusted). It imports only shared building blocks and one partner GraphQL module (graphql/partner/console.ts); components/partner-console/__tests__/import-boundary.spec.ts keeps it that way, so it can be extracted into its own portal. /p goes straight into the only firm a person may enter and otherwise lists them; a firm they are refused says why, in the API's words, with the exact next step (for the missing two-factor sign-in, a link to the profile security tab).
    • Clients: server-paged with the shared EditableTable, TablePager and SearchAndFilterBar, a 300 ms debounced search and an engagement filter. Posture cells show a number only when there is one: a metric the client withholds reads "Withheld by client", an unmeasured one "Not measured", a failed snapshot is marked stale with its last good figure and time, and none is ever shown as zero. Each engagement says what is happening, since when and what to do next (invited and waiting for the owner, the day 0 / 14 / 60 detach dates). Open client enters the workspace; Refresh posture shows the API's rate-limit message as given.
    • Onboard a client shows the band and the price (from partnerPricePreview, cents formatted as dollars) before anything is created, the API's "No rate card set" message with who to ask, and a Resend when the invitation email did not go out.
    • Team (2FA badge, add, change role, remove with the warning that it ends access in every client; admin only), Statements (issued and void, lines, a total the page re-adds; admin and billing), Templates (create and edit, offer to chosen active clients, per-client answer, withdraw).
    • The console is also reachable on its own host: on partners.securehive.ai (.app, .io, partners.localhost) the proxy serves / as /p. Nothing else about any host changes, and partners is now a reserved subdomain (never a tenant).
    • Settings › Partner access (/s/[tenant]/settings/partner-access): the engagements, who works here for the firm and what it did, withheld-area switches (privileged access, users and access, Trust Portal publishing, the four posture figures), Accept (shows and sends back the terms version, after an explicit confirmation) and Decline for an invitation, and for an active engagement the Detach flow with the 0 / 14 / 60 day explanation and the direct price, then the choice of outcome (direct licence, new partner, export). A banner on every workspace page asks the invited owner to review and answer; a firm operator sees "You are working as <you> for <partner>"; read-only and ending notices show while a detach runs. Template offers (approve or decline) are on the same page for the workspace Admin.
    • en, es and ja (partnerConsole, partnerAccess).
    • Per-client assignment (D-PTR-16). A firm has several vCISOs, each on their own clients. The list shows who works on each client with their role, the viewer's own role, and Open only when the API's canOpen is true; otherwise it says why (not assigned, engagement not live, billing members open nothing) and offers Assign to a firm admin or the client's lead, so an admin is never implied to open everything. A per-client team panel assigns, changes the role of and takes off colleagues (taking off ends access on their next request); the Team page counts each person's clients, explains that new members start with none and that billing never works in clients, and gives an admin the person-side view. Onboarding says you become the client's lead and lets you add colleagues once it exists. The operator bar's client switcher lists only clients the person can open, and the client's Partner access page shows each operator's role on it.
  • Switching workspace. The workspace a request belongs to is read from the page's URL (host, then /s/<key>, then the tenant_id cookie as a fallback) and checked by the API; the client keeps nothing else. What a page load does keep, and what went stale for a person in several workspaces: the signed-in user (user.currentTenantId, read in about 430 places), the Apollo cache (currentTenant takes no argument, so one answer is kept per client) and the cookie. Entering a workspace is now a full page navigation after the cache is cleared and the cookie pointed at it (lib/workspace-switch.ts, hooks/use-enter-workspace.ts), so a new page load rebuilds all three; none of the 430 reads is touched. TenantScopeGuard reloads the page if a client-side move ever crosses workspaces, or between the console and a workspace. The console names no workspace on its requests (resolveTenantKey returns none under /p), and AuthProvider neither writes the stored workspace into the cookie there nor shows "wrong workspace" for it.
  • Policy statement lifecycle, slices B and C (D-PSA-2..6, Screens 1–4).
    • Statements tab: tiles for partial work and its open gaps, statements needing re-assessment (with the date the wording changed) and reviews due; a banner per assessment round state (open with what is left, waiting on content digestion with what the digestion is doing, dry run, complete, closed); owners proposed from the RACI with Confirm and Confirm all; a re-assessment banner with "Show only these"; server-side filters; the verdict dimmed with the version it was given for; an Outcome column (risk, exception with its expiry, control review state, accepted with its reason, or "withheld"); bulk assessment of selected rows (no partial, all or nothing, the server's refusal shown in its words); Export for auditor (the assessment history as CSV, saying when it was cut or had withheld cells).
    • Assessment form: why it is being assessed (policy changed, review date passed, failed test, publish round), both wordings when the sentence changed, what the control already has (owner, evidence, tests, linked risks — each "withheld" in words without that module's permission), the control-review outcome addressed to the control's owner, and registering a new risk or requesting a new exception from the form with those modules' own mutations.
    • The owner's queue says every reason a statement is there: reworded (with both versions), review due or brought forward by a failed test, asked by a publish round, waiting on someone's gap (with a once-a-day Nudge), excepted until a date.
    • Standards tab: both directions of disagreement counted, each control's review state (sent, opened, reminded, or no owner) with Remind, and whose gap a control waits on.
    • Control owners get control reviews in My Assignments (?item=policy-control-review:<id>), opened when shown and resolved with a note.
    • A "Statement assessments" card on the policy-lifecycle Settings page (Verified independence, what publishing does, due days).
    • The eight new notification kinds are formatted in the viewer's language. en, es and ja.
  • Policy statement assessments (D-PSA-1..4). A statement's status is now set through an assessment form (Statements tab → Assess; My Assignments → Assess), not a status select: the verdict (now including Partially implemented), the gaps still open (each with an owner and a due date), the rationale, the artifacts of the linked control that were examined, where a not-met verdict leads (an existing registered risk, an existing exception of the policy, or accepted with a reason) and the next review date. The form shows the server's rules before Save — partial needs an open gap, done needs every gap closed, not-met needs an outcome, Verified is not offered to the statement's owner or a gap owner — and lists previous assessments. The Statements tab gains Assessed and Next review columns (review due in red) and a partially-implemented count; the owner's workspace says why a statement is in their queue (review due, gaps past their date, overdue) and lists its open gaps. Notification policy.statement.gap_assigned tells a gap's owner what is missing. en, es and ja.
  • Policy statement assessments: the form for someone without the other modules, and a table that keeps its statement. The Statements table lost its Statement column on a window 1440px wide or narrower: the fixed columns took all the width. Assessed and Next review are now one column (when, by whom, and when to look again, with "Review due since" in words once the date has passed), the table has a 1100px floor under it and scrolls sideways below that, and long titles and owner names stay inside their columns. In the assessment form, an owner with no policy permission now sees the statement's previous assessments (myPolicyStatementAssessments); a history that could not be read says so with the reason, where it said "No assessment recorded yet". The history is a server-paged table (five at a time). The risk and exception outcomes use the shared selectors (RegisteredRiskSelector, and a new shared PolicyExceptionSelector) and are offered only to someone who may record them: without registered-risks read and update, or policy-exceptions read, the option is off and says what it needs and what to do, where the picker showed "Forbidden resource". Evidence says four different things: no control, no access to artifacts, could not be loaded, none linked yet. A refused save shows the server's own words. en, es and ja.
  • Policy statement assessments: the form's Verified follows the server's independence rule over time (D-PSA-3). The statement now carries cannotVerifyUserIds — its owner, its gap owners, and whoever owned either or recorded it as implemented since it was last verified — and the form does not offer Verified to anyone in it, saying why: "Verified needs someone independent of the work. Since this statement was last verified you have owned it or one of its gaps, or recorded it as implemented, so someone else verifies it." en, es and ja.
  • ISO 13485 certification programmes (ISO_QMS_MD). The medical devices QMS profile is a known profile: worded like every ISO standard, with no statement of applicability (no control annex) and no ISO 27001 risk treatment plan export (13485 has no clause 6.1.3). The create form offers it only its own ISO13485 framework and keeps that framework from every other profile; the framework form lists the ISO 13485 practice. Management reviews on a 13485 programme are headed "Review inputs (5.6.2)" and "Review outputs (5.6.3)", and the empty state names the twelve a–l inputs; en, es and ja.

Changed

  • Security Operations packaging (D-SEC-1..3, D-TTX-7, D-TTX-9). Detection & Response and Incident Command are Security Operations Core and visible on every plan; the premium module is labelled AI SecOps and owns Threat Hunting plus the AI pages inside Detection & Response (dashboard, triage, command center, approvals, autonomy, workspace profiles, compliance posture, reports). The premium registry now matches the longest route prefix, and modules flagged lockScreenWhenUnlicensed (AI SecOps, Tabletop Exercises) show an upsell through PremiumRouteGate instead of Access Denied. Sidebar children owned by an unlicensed add-on stay listed with a lock (Exercises included); Artemis and Autonomy Policy check the AI SecOps licence directly.
  • Detection Library and Intel Feed open on Core: prompt drafting, deploy, tuning, false-positive reports and hunting from intel are marked as AI SecOps. Security-event, correlation, incident-response and telemetry widgets no longer wait for the AI licence.
  • Tabletop Exercises needs only its own licence; its pages show the add-on upsell when unlicensed.
  • In-app pricing data: the Security Operations card becomes AI SecOps (3 operators included, per-plan operator prices); new Tabletop Exercises card.
  • Pricing & packaging (D-PRC-1..7, 2026-09-25). Three public plans on /pricing and the homepage: Essentials (5-seat minimum, D-PRC-2; list prices unchanged), Professional, and Enterprise at 51+ users sold as Contact sales with no public price. The Business plan stays in pricing-data.ts as a non-public record (public: false); Enterprise carries contactSales: true and the renderers never print its numbers. getRecommendedTier() now lives in pricing-data.ts, skips non-public tiers and returns Enterprise from 51 users; the calculator shows a contact-sales panel instead of a breakdown there. Add-on cards: AI SecOps, Tabletop and Certification state Essentials / Professional terms with Enterprise "contact sales" / "included in quote" (the Business bands are the internal rate card); AI Governance states its units (25 sensors + 25 extensions, $50 / $20 per unit per year — D-PRC-3, matching the backend registry); new Executive Metrics, Application Security and Privileged Access (contact sales, BETA — D-PRC-5) cards; Agent Runtime and the AI Governance Suite bundle are public: false and off the page (D-PRC-4). The in-app Security Operations page drops the "minimum 3 operators" wording for "3 operators included" and the Professional / Enterprise columns.
  • pricing-data-parity.test.ts (vitest) pins the brief's figures and diffs every tier and module against the landing site's pricing-data.ts when that checkout is beside this one.
  • i18n: new marketing keys for the contact-sales and seat-minimum copy (en / es / ja).

Added

  • Core cap notices on the playbook list and the Security Operations integrations page (secOpsCoreUsage), and an AI SecOps operator seats card on Roles (secOpsOperatorSeats).
  • Alert intake card on the Security Operations integrations page (Core only, D-SEC-8): a switch that turns Core alert intake on or off (setSecOpsCoreAlertIntake).
  • Active switch on each playbook's page — the one place a playbook is switched on or off; on Core, switching on past 3 active is refused with the reason. Creating a playbook past the limit shows "Saved switched off" (D-SEC-5).
  • registry.test.ts (vitest) pins which module owns each Security Operations route.
  • i18n: new keys in shared, sidebar and securityoperations (en / es / ja); removed the unused tabletop.secOpsModuleRequired, tabletop.tabletopModuleRequired and securityoperations.detectionUnlicensed.

[1.22.0] - 2026-11-10

Added

  • The executive dashboard briefing, for every workspace (frontend#716, #722, #725, #735; backend#1052, #1055, #1062; admin portal#44, #46). The dashboard opens on the posture score and its trend, the decisions waiting on you, what changed since your last visit, what needs attention and what the AI agents did, above five lanes that follow the sidebar. Lanes show Cards, Stats bar or Charts (the former tile grid's widgets, D-DASH-25); the choice is saved per person, and per role under Settings › Dashboard defaults. Every licensed module has a tile; a switched-off one says since when and whom to ask.
  • Export › Choose cards… and Save as image on every card (D-DASH-24), beside Make a brief from this dashboard and Image of this page.
  • Release 1.22.0 note. The login release-notes dialog gains a 1.22.0 entry dated 2026-11-10, CURRENT_RELEASE_VERSION and package.json move to 1.22.0, and the "counted after treatment since 1.22.0" line under the risk figures now shows that date.

Changed

  • Critical and high risks are counted after treatment, on the workspace's own risk level thresholds, across risk assessments and the risk register; the posture card and the risk dashboard agree. The quick-count row is gone.
  • The tile grid and the dashboard-briefing switch are removed (D-DASH-25): everyone gets the briefing, and new viewers start on Cards (D-DASH-26).

[1.21.0] - 2026-11-06

Added

  • Bug Bounty screens (frontend#680 and the integration PR frontend#708, which carried #683, #685, #702 and #705; backend#997 and #1029). Programs, scope, the report inbox and report page, verdicts, internal notes, rewards and the Rewards tab with payment records and CSV export, the SLA pause shown to the team, the suggest-only AI triage panel and review list, and the report count. Every list is EditableTable with TablePager and server-side search and paging; strings in en, es and ja. All of it is behind the backend switches, off by default: the bug-bounty premium module per customer, bug-bounty-ai-triage and bug-bounty-reward-reminders.
  • Release 1.21.0 note. The login release-notes dialog (src/lib/release-notes.ts) gains a 1.21.0 entry for Bug Bounty and CURRENT_RELEASE_VERSION moves to 1.21.0.

[1.19.0] - 2026-09-24

Added

  • Tabletop Exercises (premium add-on tabletop-exercises, inside Security Operations): a new Exercises entry under Detection & Response, shown only when both licences are on and the viewer holds soc-exercises.read.
    • Exercise hub (/security-operations/exercises): readiness trend (median time to contain — a trajectory, never a grade), findings closed, last attestation with the InsurHive ir_plan_tested link, the next-attestation clock with the plan allowance, posture-triggered proposals from SecureHive (first run, attestation due, detection gaps, incident replay, intel match), and the exercise table with the practice / attested mode as a first-class column.
    • Scenario Composer (/exercises/new): mode chosen first and framed as a privacy promise, scenario / format / actor / duration / brief, Mentor's tailored narrative with grounding chips naming the real objects it used, dark-technique callout, and the inject timeline with branch points shown as structure while contents stay sealed. Start straight into the room.
    • Exercise Room (/exercises/[id]/room): the containment banner, elapsed and regulatory clocks (GDPR 72h, SEC 8-K, carrier), the tenant's own incident phases as a stepper, Mentor's feed with typed, timed, escapable decision points and consequence notes, the role rail with vacant seats in red, and end-to-debrief.
    • After-action (/exercises/[id]/aar): computed scorecard bars against fixed targets, findings with one-click promotion to a Sigma rule, registered risk or issue (attest permission), close / dismiss, attestation with facilitator-of-record and propagation targets, the locked evidence hash, and Mentor's narrative labelled as narrative.
  • use-tabletop.ts hooks (both licences folded into licensed), tabletop.query.ts / tabletop.mutation.ts, pure view-model helpers in src/lib/tabletop with vitest coverage (clocks, sequencing, scorecard bars, attestation rules).
  • i18n: tabletop namespace (en / es / ja) and nav.exercises.
  • Premium registry entry tabletop-exercises (no section of its own; app-sidebar hides the Exercises child without the licence and skips the placeholder for label-less add-ons).
  • New-release dialog announces 1.19.0 and Tabletop Exercises.
  • Exercise list on the hub runs on the shared EditableTable with database-side search, mode tabs and paging (useTabletopExercisesPaginated), two empty states, and proposals as feed rows.
  • Participants card in the composer: a seat per playbook role (the C-suite for an Executive sim), each a searchable picker over the workspace's members; add, clear or remove seats.
  • Live room: useRecordChannel('TABLETOP_EXERCISE') re-reads on each signal (the 10-second poll only while the channel is down), a Live pill and PresenceAvatars in the header; the room opens as a full-screen dialog over the shell with the feed styled after the dossier mockup.
  • Role-player threads under persona injects: "Reply to <persona>", the in-character replies threaded beneath, and "<persona> has moved on" after three replies (personaThread in view-model.ts).
  • Timed delivery: a switch in the composer and in the room (run permission), with a countdown to the next inject or "holding for the decision".
  • Evidence downloads on the after-action page of an attested run: Evidence (PDF) and Evidence (JSON), with the verified / mismatch line.
  • Board crisis scenario label and the board chair and investor personas; choosing the board crisis switches the composer to Executive sim.
  • The four tabletop notification kinds are localised in the notification center (notification-format.ts).
  • Grounding a viewer cannot read is withheld, not hidden: the composer, the room rail and the after-action page say how many facts were withheld and from which modules (GroundingWithheldNote), and the hub names proposals built from modules the viewer cannot open (withheldProposalKinds).
  • i18n: every new string in tabletop and notifications.kinds for en / es / ja.
  • Room actions follow the run permission: the phase stepper, Start, Next inject, End, answering a decision point and replying to a role-player are disabled for a viewer who can only watch (the backend refuses them either way). Decision-rail times parse epoch-millisecond timestamps with toDate.
  • Decision points: "N remaining" follows the first answer logged, as the backend does; option consequences arrive only after the room answers, so the type allows consequence: null.
  • Hub list requests only the fields a row renders (TabletopExerciseRowFields), and tabletop mutations refetch the hub's paginated list and readiness by operation name instead of an unused 50-row list.
  • Sidebar: soc-exercises.read admits through the Exercises child only, so a viewer holding just that permission never falls back to the section's full link list when the add-on is unlicensed.
  • Notices: a debrief notice without a findings count shows the server's text instead of "0 findings".
  • Archive on the after-action page follows the backend rule (canArchive): an attested-mode exercise waiting for its attestation offers Archive only to someone who can attest it.

[1.18.0] - 2026-09-14

Added

  • Certification Programs (premium module certification-programs): a new sidebar section under Risk & Assurance with Programs, Applicability, Nonconformities, Management Reviews, Records, Certificates and Calendar, running ISO/IEC 27001:2022 programs end to end.
    • Program hub (/certification/programs): one card per program with certificate, next obligation and readiness, a catalog card for licensed frameworks without a program, and the certification calendar beside them. The new-program dialog shows the plan quota and disables every field when no program slot is left.
    • Program cockpit (/certification/programs/[programId]): certification cycle strip, readiness KPIs, clause readiness, Annex A theme table, mandatory records, obligations and external access, with tabs for Scope & context, Requirements, Applicability (SoA), Records, Competence, Audits & milestones, Nonconformities, Management reviews and External access. The tab strip is one row that scrolls sideways on narrow windows.
    • Statement of Applicability (sidebar Applicability, with a program picker, and the cockpit tab): applicable / excluded decisions with justification, implementation status, linked risks, policy and evidence, versions with approval and exports; the workspace is one shared component.
    • Nonconformity record: grading, correction, root cause, corrective action plan and actions, certification-body deadlines, independent effectiveness verification; a timeline rail (done green, next blue, missed red) with the next surveillance or recertification visit, and linked finding, control, risks, policy, evidence, audit visit, program and last published management review.
    • Management review record: attendees (top management required), clause 9.3.2 inputs a–g computed from live data with their sources, decisions, actions, and publish.
    • Records (every program's mandatory-record register with link and unlink) and Certificates (certificates across programs with expiry, status, trust-portal state and publish).
    • Calendar with a Reminders and daily snapshots card: per-tenant switches readable by everyone and changeable by program managers.
  • Playwright project certification-ui (tests/e2e/certification/pages.spec.ts) and Vitest specs for the SoA helpers and nonconformity deadline tones.
  • certification i18n namespace in en, es and ja.
  • Sync from workflow names the people a workflow lists who are no longer active members of the tenant, and says whether the control owner or the tenant admins were asked to assign someone, or that nobody could be (inactive-assignee-notice.ts, with a Vitest spec rendered through the English messages).
  • The control test edit page opens the tab named in ?tab= (details, planning, findings, users or evidence), so the assignment email lands on Users Assigned.

Changed

  • Risk Register / My Riskspace — a risk owner without a risk role can now open their own risk and work it: the risk page admits my-riskspace.read holders (the server refuses risks they do not own), and the lifecycle and treatment panels act for owners as well as register editors. My Riskspace's Registered Risks tab shows the same Risks I own table as the review queue (owner review pending, unreviewed and overdue first), under the existing level counts; the queue and My Riskspace share one component. My Riskspace's icon rail is now full-bleed like the assignment workspace (its icons no longer sit off-centre inside the page gutter). Assessment-tab strings are translated (en/es/ja).
  • New-release dialog and sidebar announce 1.18.0 and Certification Programs.
  • Users Assigned tab strings are translated (en/es/ja).
  • Certification: the SoA workspace, the SoA version dialog and the nonconformity record say which linked data is withheld from the viewer instead of showing it as empty or zero (withheld-note.tsx). The link-record dialog shows the server's refusal, the raise-nonconformity dialog lists only the program's audits, and planning verification of a graded nonconformity waits for certification-body acceptance of its plan.
  • Publishing a certificate no longer offers a public switch: the certificate is published to the trust portal as a gated document, and a trust manager decides whether it becomes public. Without permission to create trust portal documents the Publish button stays visible but disabled, saying to ask a trust manager.

Fixed

  • Vendor questionnaire submission calls submitVendorQuestionnaireResponse, so it no longer collides with the cyber-insurance questionnaire mutation of the same name.
  • A certification-body deadline met by a submitted corrective action plan is no longer shown as overdue.

[1.17.0] - 2026-09-12

Added

  • Risk Register lifecycle: registered risks move Identified → Under review → Treating / Accepted → Mitigated → Closed, with reopen; a status tab per stage on the register list, a lifecycle stepper with only the legal moves on the risk page, and every transition in the activity timeline with actor, reason and workflow instance.
  • Review workflow by level (Risk Register → Settings → Review & cadence): a workflow template per Critical / High / Medium / Low or "No workflow — the owner reviews", plus an any-level default; the register form previews template, stages and assignees before save and names stages whose role has nobody mapped.
  • Review queue (Risk Register → Review queue): workflow stages waiting on the signed-in person, titled with the risk, and the risks they own that are unreviewed, waiting on their own review or overdue.
  • Treatment panel: Modify / Retain / Avoid / Share, treatment actions as Execution Tracking tasks (create or link), treating controls with effect, residual scoring, and a mitigate gate.
  • Assessment read-through: finding decisions (add to register, link to existing, accept as-is, not a risk) on the assessment risk table; a treatment-status card per finding with rollup, approval snapshots and a Risk Treatment Plan PDF export into the Artifact Library; submit-for-approval refused while findings are undecided.
  • Register dashboard: inherent and residual heat maps whose cells open the filtered register, status and level breakdown, owner load, and tiles for reviews overdue, unowned, stale reviews, overdue actions, expiring acceptances and open review workflows.
  • One risk settings page: Scoring & taxonomy, Review & cadence and Danger zone tabs; /risk/settings and /risk/risk-level-management redirect there.
  • Playwright spec tests/e2e/risk-register/pages.spec.ts (project risk-register-ui) for the manager, intern and admin personas.

Changed

  • Sidebar: Risk Posture → Risk Register, Risk Signals → Risk Assessments (routes unchanged); breadcrumbs, read-first and docs follow.
  • Register list: server-side search (debounced), filters and paging (page size 10/25/50/100, capped server-side), server-supplied owner and assessment filter options, Export CSV, drag-and-drop ordering kept; Delete-all moved to the settings danger zone.
  • Register form: description, initial treatment option and review cadence; owner defaults to the current user with "Assign to me"; score and level are shown as a preview and computed on the server.
  • Assignment workspace and unified task rows title registered-risk reviews with the risk id and title and open the risk page.
  • Frontend CASL: risk-activities and risk-register-admin mapped to their own subjects; reviewer-settings removed.

Fixed

  • Registered-risk reviews were labelled "Risk Assessment" and linked to risk/assessments/<id> in the workspace.
  • The register detail page crashed on a numeric review date (toDateInputValue normalises epoch and ISO values).

Security

  • Delete all risks is shown only to holders of risk-register-admin.purge (tenant Admins), and the dialog stays inert until a fresh export issues a receipt and the confirmation phrase is typed.

[1.16.0] - 2026-08-23

Added

  • Strategy v2 — "did it work", not just "did we ship it": the whole strategy surface rebuilt around three signals kept apart — delivery (did we ship it, rolled up from linked projects, goals and milestones), outcome (did the number move, from KPIs bound to the metrics engine) and health (will we make it, with a written reason). They are never averaged together, because the most valuable finding a programme produces — full delivery, flat outcome — vanishes the moment you blend them.
  • Execution overview (Security Direction → Dashboard): delivery and outcome across every strategy, side by side, each stating the number of objectives it was averaged over. A percentage with no denominator is not an answer. Below it, an honesty line (how many objectives have a KPI, how many read themselves, how many have nothing delivering them, when it was last measured) and a Needs attention list ordered by how much a human decision is required, not by health alone.
  • Strategy generator: a readiness step that counts your maturity gaps, open risks, failed control tests, coverage holes and overdue policies, recommends a mode with a reason and offers all three regardless — obligations-led, evidence-led, or continue-the-last-strategy. Nothing is written until you accept, objective by objective. Non-goals are treated as absolute. Truncation is always reported.
  • KPI binding by description: describe what you want to move — "cut open critical risks to under 5" — and get a real metric with the current value as baseline and direction derived from baseline→target. Describe something nothing measures and it says so, with a reason, and offers no binding.
  • What the money bought: budget and spend rolled up from the projects actually delivering an objective, reported one row per currency — currencies are never added together — with spend beside outcome and never divided into it. Plus Spent, not moved: objectives that consumed budget without moving their number, highest spend first.
  • Roll-forward: name any next period (annual, multi-year, rolling) and every objective is classified from where it actually finished — carried, restated or closed. Achieved work is left behind with its reason and a tick to bring it forward anyway.
  • Security charter drafter and completeness review: clauses drafted from your own records, each marked cited / unverified citation / your own choice, and scored against eight published rubric criteria with stated weights. Suggested wording is offered to copy, never applied.
  • Findings inbox (Security Direction → Findings): weekly snapshots compared period-on-period, material movements routed to whoever actually owns the work — derived from what is delivering the objective — with a portfolio roll-up for the CISO and a plain sentence on every card saying why it reached you.
  • Budget scenario modelling: which objectives would fall below money already committed, stated as arithmetic, with what it cannot tell you printed alongside every result rather than behind a link.
  • Board pack: the strategy slide quotes delivery and outcome separately with denominators, delivered-but-unmoved objectives get their own named slide, and a Check against source panel above the deck shows the live figures every number should trace back to.

Changed

  • Security Direction → Dashboard now opens the execution overview. It used to open the strategy list, which is not a dashboard; the list keeps its own entry.
  • Strategy objectives, KPIs, evidence and delivery links are first-class records with codes, replacing the JSON blob that could not be measured, linked or rolled forward.
  • The business unit hierarchy is assembled in the browser from parentId instead of requested as five hand-unrolled levels of children.
  • New-release dialog and sidebar announce 1.16.0 and Strategy v2.

Fixed

  • The business units page returned an error for every tenant. Its query priced at 666,660 against a complexity limit of 3,000 and was rejected outright, so the organisation hierarchy never rendered. (SecureHiveAi/frontend#162)
  • Query complexity was calibrated against documents no client ever sends. Apollo injects __typename into every selection set and the validator multiplies by ten per list level, so one injected field three lists deep cost 1,000. Several operations were rejected in normal use — maturity frameworks, trust questionnaires, framework cloning. (SecureHiveAi/backend#191)
  • Accepting a generated strategy could half-create it: a KPI whose target sat on the wrong side of its baseline failed partway through, leaving some objectives created and others missing. Every KPI is now reconciled before the first write, so an accept can only skip a KPI — with a reason naming it — never fail on one.
  • Subtree height in the business unit tree was computed with a filter per node, O(n²) over the tree; it is now a single pass.

Security

  • Weekly assurance ships flag-gated and dry-run by default, resolves its permission through the Autonomy Policy Grid on every run, and records every pass in the agent ledger. A held run says plainly that it was held and why.

[1.15.0] - 2026-08-19

Added

  • Application Security (new premium module): application registry with GitHub import (server-side repo search — a 3,000-repo estate browses like a ten-repo one), findings, AI threat modeling (tenant-configured model, every run in the agent audit trail), SDLC gates, and workspace-scoped access.
  • GitHub Advanced Security integration: code scanning + Dependabot + secret scanning in one queue-backed sync that never blocks the app; full alert lifecycle (fixed→resolved, dismissed→accepted, recurrences reopen — analyst decisions outrank the scanner and survive re-syncs); webhooks for real-time updates; per-tenant sync scope (linked / selected / topic / all, skip-archived) and schedule (interval or cron); per-application sync; connect-existing-installation claim flow.
  • SARIF ingest endpoint: any scanner that emits SARIF 2.1.0 (Trivy, Semgrep, Gitleaks, OWASP ZAP, …) pushes findings from CI with a per-tenant rotatable show-once token; uploads are complete snapshots (absent findings auto-resolve, suppressions arrive as accepted risk, recurrences reopen), unmatched repositories are rejected explicitly, and a Security Scanners settings card ships token management, upload health, and copy-paste CI snippets.
  • Findings triage: faceted source/severity/status/repository filters with live counts, GitHub deep links, permission-gated status changes.
  • Executive surfaces: GitHub Security Findings widget on the CISO dashboard (severity donut, per-source tiles, top applications, never-synced vs all-clear empty states); Application Security as a weighted category in the Overall Risk Posture Score (counts only when licensed and populated); three Executive Metrics — findings open / ingested / MTTR — filterable down to a single repository, with searchable repository/application filter values in the metric picker.
  • Coming next on integrations: Snyk, GitLab Ultimate, Semgrep, and SonarCloud connectors are planned over the next few weeks, feeding the same findings queue and lifecycle.

Changed

  • Enabling a premium module for a tenant now auto-provisions its default role permissions — a newly licensed module's navigation appears immediately instead of silently vanishing.
  • Licensed modules a user's role cannot access show a clear "no access — ask your administrator" sidebar state instead of being hidden.
  • The risk-score explainer dialog is wider and includes the new Application Security category.
  • New-release dialog and sidebar now announce 1.15.0 and the Application Security module.

Fixed

  • Dashboard Save failures in Executive Metrics are now surfaced with the exact validation issues (previously a rejected save looked like a success); metric panels added from the picker save cleanly.
  • Finding close-timestamps are stamped from GitHub's own fixed/dismissed times, so open-over-time series and MTTR are accurate.

Security

  • All GitHub App credentials are tenant-key encrypted; webhook and install-state signatures are verified with constant-time comparison and expiring, domain-separated tokens; every AI threat-extraction run is audited; findings access respects per-application workspace membership.

[1.11.0] - 2026-07-13

Added

  • AI SOC Layer (Aegis) — Beta (Security Operations → Detection & Response): an AI layer that works the alert queue with you — Alert Triage, Approval Inbox, Autonomy Policy, Command Center, Workspace Profiles, and an AI Case Feed. The new AI SOC pages are labelled Beta while we harden them with design partners.
  • Autonomy Policy Grid: one authority for how far Aegis may act per risk class (Suppress alert, Open incident, Auto-investigate, Respond), from Human-gated to Auto; Respond carries a non-degradable gated floor.
  • AI Case Feed: a plain-English running narrative on every incident (what Aegis observed, concluded, and did).
  • Connector credential health: per-stack token-expiry countdown and a classified 401 (expired / wrong-stack / decrypt-drift / capability) on Test Connection; a scheduled monitor warns tenant admins before expiry and on failure, with a rolling check history for intermittent faults, plus an on-demand "run health check now".
  • ITSM incident sync (Beta): push incidents to Jira / ServiceNow with a configurable outbound status map, per-integration dark-launch flag; the handoff policy (auto-create + minimum severity) is editable in the integration form.

Changed

  • Test Connection reports each Splunk stack (main / ES) independently, with a token-expiry countdown and a specific remedy on failure; repeated manual tests build a rolling health history.
  • New-release dialog and sidebar now announce 1.11.0 and the AI SOC beta.

Fixed

  • Autonomy Policy grid now renders the Respond risk class (previously missing from the grid).
  • Saving a playbook with duplicate incident types returns a friendly 409 conflict instead of a 500.
  • Incident-linked hunts that fail, abort, or are cancelled now write an entry to the incident's Case Feed instead of ending silently.
  • Alert-suppression ceiling is clamped so High/Critical alerts can never be suppressed (max suppressible severity is MEDIUM).

Security

  • Everything Aegis proposes or does is policy-gated, tenant-isolated, cost-bounded, and fully audited; high-risk response actions never run without a human unless the Autonomy Policy explicitly allows it.
  • Connector credential-health probes are read-only; the classifier distinguishes a stack-side token failure from a SecureHive-side tenant-key (decrypt-envelope) issue, so support never asks a customer to re-mint a healthy token.

[1.10.0] - 2026-07-02

Added

  • Artemis — AI Threat-Hunt Agent (Security Operations module): an autonomous, hypothesis-driven, human-in-the-loop hunt agent under Threat Hunting → Artemis. Describe an objective in plain English; Artemis runs the 5C loop (Contextualize → Compose → Collect → Correlate → Conclude), executes read-only against your sources, and concludes with findings, a report, proposed Sigma/YARA detections, and ATT&CK coverage. Nothing runs until you approve it.
  • Adaptive (agentic) mode alongside the guided plan: a reasoning loop that chooses each next read-only action from live results (no fixed script), with customer-selectable autonomy — Approve-once · pause on high-risk, Gate every action, or Full autonomous (budget-bounded; available to tenants using their own AI provider or granted the entitlement).
  • AI-authored, self-correcting SIEM search (siem.search): translates a natural-language intent into native SPL/KQL/FQL via your AI provider, runs it read-only, and refines against results — with a read-only guard blocking mutating commands.
  • Scheduled hunts: save an objective to run on a cron cadence (e.g. a nightly brute-force sweep); create/enable/disable/delete from the cockpit.
  • Run controls: Stop, Pause, and Resume an in-flight hunt from the cockpit header.
  • Bring-your-own threat-intel: VirusTotal, AbuseIPDB, and Shodan enrichment use per-tenant, encrypted API keys (Settings → Integrations → Security Operations).
  • Analyst cockpit: evidence browser (native query + result + raw JSON, copy-to-clipboard), confirm/dismiss findings, report sign-off, ATT&CK coverage chips, and one-click follow-up hunts.

Changed

  • Inter-step data flow: steps that depend on an earlier result resolve the concrete value from prior evidence before running; the planner is given each tool's exact input schema so chained hunts use real values, not placeholders.
  • Reliability: every external call (embeddings, LLM, SIEM) is time-bounded so a hunt can't hang; a failed run reports the reason instead of stalling silently; hunt jobs run at higher concurrency so one slow hunt can't block others.
  • Readability: step descriptions wrap to two lines and each evidence card has a copy-raw-result control.

Fixed

  • Hunts against live SIEM data no longer fail to persist: results carrying timestamps, large payloads, or numeric IDs are sanitized before storage.
  • Enrichment steps return a clear "could not resolve a concrete indicator" message instead of a confusing upstream error when a value can't be derived from earlier steps.

Security

  • Artemis is read-only against all external systems (mutating tools are hard-blocked); every action is human-gated by policy (a prompt can make a step stricter but never weaker), tenant-isolated, cost-capped, and fully audited.
  • Full autonomous mode is eligibility-gated — available only to tenants using their own AI provider or explicitly granted the entitlement — and still bounded by the read-only invariant, per-hunt budget, and audit trail.

[1.8.0] - 2026-06-14

Added

  • Multi-language support (Beta): the app, landing page, and pricing are now localized in English, Japanese (日本語), and Spanish (Español). Translations are in beta — please report wrong words or translations so we can fix them.
  • Language switcher on public/marketing pages, plus a per-user language preference in user settings (persisted via the NEXT_LOCALE cookie).
  • "What's New" release dialog now announces the multi-language beta and invites translation feedback.

Changed

  • Locale resolution: public pages auto-detect the visitor's browser language (Accept-Language); the authenticated app follows the saved per-user preference and defaults to English.
  • Brand names, security standards, acronyms, and identifiers remain in English by design across all locales.

Notes

  • Arabic and right-to-left (RTL) layout are planned but not yet enabled.

[1.6.0] - 2026-06-09

Added

  • Security Operations module (premium add-on): Threat Hunting, Detection & Response, and Incident Command unified under one separately licensed module.
  • Platform Admin: per-tenant Security Operations toggle on the customer features page.

Changed

  • Premium-module registry generalized to support modules spanning multiple top-level routes (routeSegments); sidebar and route gating now license all three Security Operations sections together via isSidebarLabelLicensed.

[1.4.0] - 2026-04-16

Added

  • Security Portfolio: New tenant area under /s/[tenant]/security-portfolio—technology register, detail pages, capability catalog (/security-portfolio/capabilities), CSV export, and portfolio audit log consumption in the UI.
  • GraphQL hooks: security-portfolio queries and mutations wired for list, detail, capabilities, coverage, controls, artifacts, and AI suggestion flows where enabled.
  • RBAC: UI gated by security-portfolio permissions (read, create, update, delete, manage).

Changed

Fixed

Security

1.3.0 - 2026-03-14

Added

  • Splunk ES integration: Connect Splunk Enterprise Security to Mission Control and incident management. Notable events, Mission Control alerts and incidents; create Secure Hive incidents from Splunk alerts. ES instance config in integration settings; Urgency (5 levels) with theme-aligned tags.
  • Microsoft Sentinel integration: Client credentials (tenant, app, secret) with backend token management; ARM Log Analytics query for all regions. View incidents and alerts, create Secure Hive incidents from alerts, Open in Azure in actions.
  • Mission Control (SIEM-agnostic): Severity (Sentinel, 4 levels) and Urgency (Splunk, 5 levels) with theme-aligned tags. Compact action buttons with tooltips; Open in Azure in actions column.
  • Incident and alert UX: Enriched alert details on incident page (description, created/updated, product, alert ID). Azure portal incident URL in actions. Create Secure Hive incident from Sentinel or Splunk alerts.
  • Presentations: Gamma integration and a beautiful AI dialog to create tailored board presentations from customer data for stakeholder and board reporting. Export to Gamma; guided AI flow for focus, audience, and narrative style.

Changed

Fixed

Security

1.2.0 - 2026-02-16

Added

  • Incident Management: Full incident lifecycle—create, triage, assign, and track security incidents. Streamlined workflows to coordinate response and keep stakeholders informed. Link incidents to maturity capability assessments and other modules.
  • Maturity Model: New maturity model module with framework-based assessments. CIS Controls framework support: license from marketplace, assess capabilities by domain, track maturity levels and applicability (IG1/IG2/IG3). Questionnaire-style assessment UI with search and collapsible domains. Other frameworks (CMMC, NIST CSF, ISO 27001) coming soon—unified schema and CSV import ready.

Changed

Fixed

Security

1.1.0 - 2026-03-06

Added

  • Platform Evaluation module: compare candidate security platforms (e.g. EDR, SIEM) with weighted criteria
  • Evaluators and observers; score matrix and weighted totals
  • Criteria library, add-from-library, and optional AI-suggested criteria
  • Criteria proposal workflow (request, approve, reject)
  • Complete evaluation with selected platform and rationale; link to strategy project

Changed

Fixed

Security

1.0.0 - 2026-01-18

Added

  • Modular platform: Governance, Strategy, Risk, Audit, Policy Lifecycle, Vendor Risk, Trust Portals, Incident Response
  • Unified view of security posture across modules; strategy-to-execution (Charter → Strategy → Goals → Projects → Tasks)
  • Comprehensive CISO observability: risk posture score, executive and board dashboards, real-time analytics, automated risk scoring and compliance metrics
  • Unified approval workflow system for Policy, Strategy, Charter, and Issues with automated assignments, notifications, and task routing; workflow builder and tenant/instance-level configuration
  • Feature flags context and hooks for React/Next.js (gradual rollouts, A/B testing)
  • FeatureGate component for conditional rendering based on feature flag state
  • Version display components (aligned with release documentation)
  • Changelog page for displaying release history within the application

Changed

  • Performance optimizations across the platform
  • Improved error handling and user feedback
  • Release documentation and support materials

Fixed

  • Stability and edge-case fixes ahead of beta

Security