Changelog

Improvements

• Tooling: The community map build now imports the Node path helper using the node: prefix recommended for modern Node and bundlers, reducing ambiguity during builds.

What's New

• Two PDFs when it helps: For supported collections, scheduled jobs can attach an analytics-style PDF (with charts) in addition to the standard tabular export—so recipients see trends at a glance and still have raw rows.
• Clearer attachments: Email attachments use consistent, human-readable names that reflect the report type and date range.
• Automated quality: The worker now ships unit tests around export routing, PDF helpers, and key analytics paths so regressions are caught earlier.

Improvements

• Email copy: Messages reflect when multiple attachments are included.

What's New

• Patrol analytics exports: Patrol PDF and HTML reports surface SLA-oriented metrics, checkpoint detail, and clearer summaries for management reviews.
• Field access exports: Dedicated field-access analytics and report flows (including licence-disc style summaries where applicable) so operations teams can export the right dataset without mixing contexts.
• Live map: The field-access map shows whether a guard is effectively online for live tracking, with actions for operators when someone needs to turn location sharing back on.
• Site settings: Sites can store an optional patrol completion threshold used when configuring schedules and expectations.

Improvements

• Safety routing client: Small updates to support batch scoring helpers used by enterprise flows.
• Export tooling: Shared HTML-to-PDF utilities reduce duplicate report code across patrol and field-access exports.

What's New

• Location prompt handling: If you receive a control-room prompt to share location, the app checks your grant state and triggers the idle location heartbeat path so GPS updates resume in line with your permissions.

Improvements

• Faster recovery: Reduces cases where the map looks stale because background location needed a nudge after a notification.

What's New

• Road-aware routing: Where road network data is loaded for South Africa, routing can prefer real road geometry after the safest hex corridor is found—better paths when the data is present.
• Operations pipeline: Documented and scripted steps to load road data into the platform database on Linux-style hosts (with automation-friendly containers where needed).
• Consistent errors: API failures return a predictable JSON shape (human message, HTTP status, machine-readable code, and request correlation when available)—easier for partners to log and retry.
• Discovery: A small metadata response describes what the running instance supports (version, capabilities, documentation links) so integrations can self-check after deploy.

Improvements

• Wrong HTTP method: Clients using GET on routes that expect POST receive a clear 405 with guidance instead of a confusing 404.
• Rate limiting: Throttled responses use the same error format as the rest of the API.
• Documentation: Expanded public docs for integration, collections, and operations (including how road data is produced).

What's New

• Location nudge: Control room staff can send a push prompt so a guard is reminded to open the app and share location when visibility is needed.
• Per-site patrol expectations: Organisations can set a minimum checkpoint completion target for a site (optional), so reporting and exports align with how strict each site should be.
• Operational awareness: Last-known guard positions are retained longer before expiring, so maps and dashboards remain useful during longer shifts or delayed updates.

Improvements

• Safer scheduling identity: Background schedule processing now handles organisation identifiers more defensively when resolving who initiated a job.

What's New

• Startup errors: If the app can’t reach the server during sign-in, you see a plain-language explanation, what to try (network, signal, switching networks), and a Try again action instead of a silent failure.
• Font loading: If bundled fonts fail to load, you get guidance and Try again (or close and reopen the app) instead of a blank or crashed screen.
• Release: App version 1.1.0 (package and store metadata).

Reliability

• Session check distinguishes connectivity problems from an invalid or expired session, so offline users aren’t left guessing.

Note

There were no new commits on origin/main for wolf-community-frontend on 2026-03-25. The active ship for this period is 2026-03-23 — Live map parity, clearer loading, and safer public layers.

What's New

• Live tab (PR #75): patrol-field-access gains a Live experience with a staff map modal, trail layers, and Sonar-friendly refactors on the new UI.
• Live map + heartbeat (PR #76): Field access live map and staff heartbeat socket integration connect the enterprise map to live positions and heartbeat signals from the platform stack.

Engineering reference

• Merge commits on main: 305d9f4 (PR #75), 77e7ada (PR #76).
• Notable commits: 65d70c2 (Live tab / staff map / trails), 9827b2f (field access live map + staff heartbeat socket).
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-enterprise-frontend-full.diff (cumulative range as for 2026-03-24 — ef702d9^..origin/main).

Follow-up (live tracking map)

• LiveTrackingTab: Persistent Map<staffId, Marker> markers with setLngLat updates, initial fit once (not on every GPS tick), Fit all control, and map teardown cleanup that snapshots the markers map at effect start for eslint react-hooks/exhaustive-deps compliance.

Follow-up (field access connection resilience)

• Live map: You can see when the field access map is offline or reconnecting, and it recovers more reliably when the browser tab was in the background or the network dropped.
• Live tracking: Patrol-events live socket behavior matches the same reconnect-on-focus pattern so the operations view does not silently stall.

Engineering reference (field access resilience ship)

• Commit on feat/patrol-field-governance-ux-audit: 7594bef (patrol events socket tuning, field access live map connection state, LiveTrackingTab wiring).

What's New

• Field-access gates: Stats-oriented home, scope copy, and Sonar-friendly UI refinements on top of the new grant model (d1c2f55).
• Idle heartbeat: Idle GPS heartbeat and GrantCapabilitiesPayload alignment so the client advertises and receives grant capability state consistently with the API (d1d8723).

Bug Fixes

• Tabs enforcement: Routes and sites tabs now enforce field access so users without the right grant cannot reach configuration surfaces (e7ef414).

Engineering reference

• Commits on master: d1c2f55, e7ef414, d1d8723 (2026-03-25).
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-patrol-app-full.diff (cumulative from 78529d0^..origin/master).

Follow-up (route creation and maps)

• Route create: After a successful create, success haptics, navigation to /route/{id}?created=1, and a route detail banner confirming the saved name.
• MapViewSafe / web: Namespace import for MapLibre (non-deprecated default export), MapMarkerVariant including dot, readonly marker props, replaceAll for stable marker ids, and MapMarkerVariant type alias on web.
• Schedules API: CreatePatrolScheduleInput extended with optional useTimeRange, startTime, endTime for device schedule creation.
• Dependencies: react-native-svg added as the required peer for Lucide icons so the native build and Expo Doctor checks stay healthy on custom dev clients.

Follow-up (patrol assignments and realtime resilience)

• Assignments tab: Browse, search, and filter patrol assignments; open an assignment to edit details, pause or resume, or remove assignments you manage from the app.
• Create patrol: Wizard split into clearer steps, with priority and description so new assignments match how your team works.
• Access prompts: When your permissions need a refresh, the app surfaces that clearly instead of leaving you stuck.
• Realtime: Patrol-related live updates reconnect more reliably after network blips or long backgrounding.

Engineering reference (patrol assignments ship)

• Commit on master: ca96e96 (assignments tab, edit flow, patrol API client, socket tuning).

Follow-up (four-tab navigation and access hardening)

• Simpler navigation: Home, Places (routes and sites in one place), Patrol (assignments and schedules together), and More (field access and account) replace a long row of separate tabs.
• Stricter lock when you lack access: The app keeps you in field access and More until you are approved, with a full-screen prompt so you are not browsing sensitive lists without permission.
• Smoother errors and refresh: Clearer error messages when something fails, and more reliable pull-to-refresh behaviour on field access and lists.

Engineering reference (patrol nav and lock ship)

• Branch: feat/patrol-tabs-security-nav — includes tab restructure, grant overlay ordering, pathname guard, hidden legacy assignments redirect, and follow-up Sonar cleanups.

Note

There were no new commits on origin/main for wolf-safety-router (Go) on 2026-03-25. The active ship for this period is 2026-03-23 — Safety router (Go) — public hex auth, CORS, and enterprise rate limits.

What's New

• Patrol stats and gating: Patrol statistics endpoints and gating for upcoming patrol-related work, with Sonar-friendly refactors on the device-auth side.
• Patrol events: Breadcrumbs on patrol-event flows improve traceability across related records.
• Live operations: Staff live positions, a heartbeat API, and grant tracking mode support real-time awareness of who is in the field and how grant state is applied.

Engineering reference

• First-parent merges on main: PR #56 and #57 (b4aae8b, b4e70cc) on feat/patrol-field-audit-device-writes.
• Notable commits: bd1ab73 (patrol-stats / upcoming gating / patrol-events breadcrumbs), 7bc0a3a (staff live positions, heartbeat API, grant tracking mode).
• Schema and modules (from diff summary): new staff-live-position schema, expanded patrol schedule and related patrol route/site schemas, supporting staff repository and middleware updates.
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-server-full.diff (cumulative from 3a0ad45^..origin/main).

Follow-up (device patrol schedules and assignments)

• Device schedule create (POST /device-auth/patrol-schedules): Safer parsing of startDatetime / endDatetime (string, number, or Date only — no arbitrary object stringification), optional useTimeRange with startTime / endTime (HH:MM), and refactors (parseDeviceScheduleCreateBody, parseScheduleTimeRange) to satisfy lint complexity rules.
• Device patrol create (POST /device-auth/patrols): For own_device_created grants, the route must belong to the requesting device user (createdByStaffId), enforced via findPatrolRouteForDeviceCreate.

Follow-up (device patrol assignment updates and live location)

• Assignments from the field: Staff can update assignment details, toggle active state, and retire assignments from the patrol app where permitted, with audit history for accountability.
• Live location: Processing path for device-reported positions is tightened so maps and operations views stay aligned with what the field app sends.

Engineering reference (device patrol updates ship)

• Commit on feat/patrol-field-audit-device-writes: 5d1cf92 (device patrol updates, audit, GPS socket handler extraction, live-position wiring).

Follow-up (field access enforcement by default)

• Device field configuration: The platform now expects approved field access on patrol devices by default before staff can use organisation patrol configuration and related data through the device API.
• Legacy deployments: Organisations that still need an older “open” device mode should align with their deployment team to opt in explicitly so behaviour stays intentional.

Engineering reference (secure-default enforcement)

• Branch: feat/patrol-grant-secure-default — secure-by-default patrol grant enforcement with documented legacy opt-out for operators.

Note

wolf-whatsapp-server had no commits on origin/main during 2026-03-22 through 2026-03-25 (local clone after git fetch). There is no git diff artifact for this product in internal/git-audits/2026-03-22-to-03-25/.

Patrol and messaging–related changes in the same calendar window were delivered through wolf-server, wolf-patrol-app, wolf-enterprise-frontend, and wolf-changelog entries for those products.

Note

There were no new commits on origin/main for wolf-community-frontend on 2026-03-24. The active ship for this period is 2026-03-23 — Live map parity, clearer loading, and safer public layers.

What's New

• Patrol Field Access & Audit hub: Dedicated Patrol Field Access experience with navigation into the hub, initial access-hub UI work, and ESLint dependency alignment for the new surface.
• Stripe billing (PR #69): feature/stripe-integration-v2 is merged to main—checkout drafting, sales-oriented billing flows, net billing change visibility in the checkout builder, patrol management display pricing update, and the Billing workspace UI (with feature gating). Lockfile and Vercel optional dependencies (Rollup, SWC Linux bindings) support npm ci on CI and production builds.
• Governance UX (PR #71): Staff names, clearer activity copy, and dialogs for deny, revoke, and reject paths on patrol-field governance.
• Realtime grants (PR #72): Grant requests and device acknowledgement surface as toasts so operators see the handshake in real time.
• Field updates (PR #73): Live updates and UI polish across patrol field access views.
• Live tracking (PR #74): Live tracking tab and patrol-event:location socket integration for location streams on patrol events.

Improvements

• Dependencies: npm audit fix passes; @swc/core pinned to 1.15.21 for lockfile parity with npm ci.
• CI / Vercel: Lockfile regeneration with npm 11, Rollup and SWC native bindings for Linux builds, and workflow alignment for Node 24 + cached npm ci.
• Integration hygiene: Staging merges and a short revert / reapply cycle on feature/patrol-field-access-hub align history while keeping patrol-field and billing work consistent (see merge commits below).

Bug Fixes

• Dependencies: Pinned toolchain packages resolve optional dependency drift between local and Vercel environments.

Engineering reference

• Merge commits on main (selected): 4f450f2 (PR #69 — Stripe), 9894dcc (PR #70 — patrol-field-access-hub), 0f2b158 (PR #71), 20c08c6 (PR #72), 62bb14c (PR #73), 588309b (PR #74). Supporting merges: a872380, 027366e, 0e168e1, 586b1c0 (staging alignment / revert-reapply choreography).
• Notable non-merge commits: 56b2d05, e16087a, efaa106, 954f81c, 7b66aa4, 64399a2.
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-enterprise-frontend-full.diff (range ef702d9^..origin/main — first first-parent commit on or after 2026-03-22 through tip).

What's New

• Mobile UX (PR #1 — feature/patrol-field-mobile-ux): Faster map GPS, location prompt behaviour on tabs, dashboard split, Expo / tsconfig ergonomics, and related mobile tab work merged to master (78529d0, c6c5f29).
• UX clarity (PR #2 — feat/patrol-field-ux-clarity): Documentation clarifying direct site/route writes versus the optional proposal queue; browse sites and routes, push registration, and assignments (525af68, 442fd5f).
• Field access: Capabilities API surface, grant version headers, schedules and access tabs; field-access gate, Routes tab, site/route editing, GPS and API client updates (5f68aee).
• Home and reliability: Upcoming patrol home emphasis, grant lockout messaging, offline UX, and push entrypoint wiring (0e09a3b).

Improvements

• API clients: Expanded patrol-event, route, site, schedules, staff, and capabilities modules align the app with server grant and field-access APIs.

Engineering reference

• Repository: Community-Wolf-Limited/patrol-helper — default branch master (not main).
• Merge commits: 78529d0 (PR #1), 525af68 (PR #2), 442fd5f (merge UX branch to master).
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-patrol-app-full.diff (range 78529d0^..origin/master — from parent of PR #1 merge through tip).

Note

There were no new commits on origin/main for wolf-safety-router (Go) on 2026-03-24. The active ship for this period is 2026-03-23 — Safety router (Go) — public hex auth, CORS, and enterprise rate limits.

What's New

• Patrol field grants (device auth): APIs and socket flows for field grants, proposals, and governance, with Clerk token verification on the socket path so device and staff sessions align with org policy.
• Governance and audit: Patrol-field governance streams and quality hardening around device-scoped writes; audit logging covers device-created sites, routes, and patrol events (including proposal flows).
• Operations: New DEVICE_PATROL_GRANT_REQUIRED environment flag is documented for deployments that must enforce grant checks for field devices.
• Realtime and assignments: Patrol-field push, grant-related socket events, and staff assignment APIs wire the mobile and enterprise surfaces to the same grant state.
• Device writes and notifications: Audit of device writes, configuration notifications, and related Sonar-driven cleanups keep patrol configuration changes observable.
• Mobile home: Upcoming patrol events for the mobile home experience, including Expo push behaviour when a token is not yet available.

Improvements

• Socket stability: Clerk verifyToken errors (for example expired JWTs) are caught so they cannot terminate the server process—clients get a controlled failure instead of a hard crash.

Bug Fixes

• Sockets: Expired or invalid Clerk tokens on the patrol/device socket path no longer risk taking down the whole Node process.

Engineering reference

• First-parent merges on main (UTC+2 window): PR #52–#55 from feat/patrol-field-audit-device-writes (merge commits 4ef4842, 02dbe0d, 790355a, b9520be on the integration train).
• Notable commits: 9b76a66 (device-auth grants / governance API / Clerk socket verify), a74a6b6 (patrol-field governance and audit streams), aa2bdbd (socket Clerk error handling), 6674cb9 / 832b94d (audit log + env docs), c42e916, c122f40, 0f7bb7c (push, device write audit, upcoming events + push).
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-server-full.diff (cumulative from 3a0ad45^..origin/main).

What's New

• Enterprise-aligned map: Layer behaviour, safety overlays, and hex colouring are brought in line with the enterprise live map so community and business views feel consistent.
• National crime insight: Large-area crime hex layers load with clearer progress and work reliably in the browser by loading through the app’s own infrastructure instead of calling third-party origins directly.
• Loading feedback: While heavy layers are fetching, the map explains what is still loading and can hold unrelated toggles steady so you are not left guessing whether the map is stuck.

Improvements

• Layer stack: The order of map layers is rationalised so overlays (reports, heat-style views, points of interest) read predictably when several are on at once.
• Production networking: In production, map and data requests use the app’s hosted paths so sessions are less likely to hit browser security limits during normal browsing.
• Zoom and pan: Maximum zoom is aligned with the enterprise map so report detail and neighbourhood context match expectations.
• Tooltips: Tooltips on map controls render above the map and avoid clipping or fighting with the map surface.
• Defaults: Some experimental or redundant layer options are removed or hidden so the default experience stays focused on what most residents use.

Bug Fixes

• Map configuration: Removed invalid assignments to read-only map SDK settings when tightening telemetry handling.
• Environment handling: Local development and production base URLs for safety and map services are resolved more predictably.

Engineering reference

• Window: All main activity for this release landed on 2026-03-23 (no further merges on main on 2026-03-24 or 2026-03-25 in this audit).
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-community-frontend-full.diff (range c2e3ef4^..origin/main at time of export).

What's New

• Automatic full access for organisations: Under Admin → Organisations, creating or updating an organisation applies every configured feature-access role in one step—no checklist—so new clients start with full capability. The dialog explains this and links to Feature Access if you need to narrow access for specific organisations later.

Improvements

• Live map POI loading: If progressive POI loading cannot complete (for example when the browser receives an unexpected response), the map falls back to a standard data request so points of interest can still appear instead of failing silently.
• POI requests: POI streaming uses the configured API base consistently and treats non-data responses more safely so the map does not misread errors as empty data.

Bug Fixes

• Admin organisations: Save stays disabled until roles have loaded so create/update always sends a complete role set; server errors are logged and surfaced with a clear toast.

What's New

• Stack: These changes ship in wolf-safety-router, the Go service behind public hex/config traffic and related routing (not the Node wolf-server).

Improvements

• Public hex reads: GET /hexes/all and GET /hexes/config are exempt from auth where appropriate so the public map can load configuration and hex data without a session, while authenticated routes keep normal rules.
• CORS for public data: Public hex responses only allow trusted Community Wolf web origins (e.g. *.communitywolf.com) instead of wide-open browser origins.
• Enterprise vs anonymous: Rate limiting distinguishes enterprise clients from anonymous traffic so dashboards are less likely to be throttled under normal load.
• Refactors: Wildcard CORS handling and related auth paths were tightened with lint fixes in the Go handlers and middleware.

Bug Fixes

• Auth middleware: Public configuration and hex requests are no longer blocked incorrectly by authentication checks.
• Lint and quality: Small cleanups in routing and CORS code paths for maintainability.

Engineering reference

• Window: All main activity for this release landed on 2026-03-23 (no further merges on main on 2026-03-24 or 2026-03-25 in this audit).
• Commits (chronological): 95b7d9d — exempt public hex GETs from auth; 28c0f7b — restrict public hex CORS; 9dd60a3 — enterprise rate limits + CORS/auth follow-up.
• Full diff artifact (Go tree): internal/git-audits/2026-03-22-to-03-25/wolf-safety-router-full.diff (range 95b7d9d^..origin/main at time of export).

What's New

• POI streaming: The /poi/stream path is tuned for NDJSON clients: responses honour Accept negotiation used by browsers and proxies, and compression is skipped on this route so chunks are not buffered in ways that break progressive loading.

Engineering reference

• Merge base for 2026-03-22 → 03-25 audit: cumulative diff on main from parent of 3a0ad45 through tip includes this commit (3a0ad45 — fix(poi): CORS Accept for NDJSON stream, skip compression on /poi/stream).
• Full diff artifact: internal/git-audits/2026-03-22-to-03-25/wolf-server-full.diff (same range as the audit script: 3a0ad45^..origin/main at time of export).

What's New

• Named layer loading: When multiple map layers are still loading, the UI calls out which layers are in flight (places, reports, crime hex, national hex, precincts) instead of a single generic spinner.
• Heavy vs light loading: Large jobs—precinct boundaries or national crime hex—use a dedicated loading strip with status text; lighter combinations use a compact banner so the map stays readable during mixed load states.
• Crime hex viewport: Shared helpers keep crime-density views aligned with what is on screen, reducing mismatches when you pan or zoom during analysis.
• Map filter state: Filter state on the live map is extended so layer and analysis choices stay coherent when switching between dense and sparse viewports.

Improvements

• Live map layer stack: On /map, data layers follow a consistent bottom-to-top order—national crime hex at the base, then viewport crime hex and precincts, then heatmap and hotspot density, then report markers, with POI symbols on top—so overlays read predictably when multiple layers are on.
• Precinct streaming progress: As precinct boundaries stream in, you see loaded versus total counts and a progress bar, and map geometry updates are batched so the interface stays responsive while batches apply.
• National hex progress: National hex loads show how many cells are fetched, how many are on the map, and optional status text when the job is still warming up—so analysts know the difference between “still downloading” and “rendering.”
• Smoother precinct paint: Precinct batches are scheduled into animation frames before hitting the map, reducing jank when many boundaries arrive in quick succession.
• Layer-name branch merges: Follow-up merges complete the human-friendly layer and precinct refactor train, keeping map pages, crime-hex handling, and safety routing in sync with the March release series.
• Safety routing: The safety-routing page is trimmed so the main map and controls stay in focus without extra chrome.
• Precinct and hex clients: Calls that load precinct and cached boundary data are tightened so repeat visits reuse work and fail more gracefully when data is still warming up.
• Staff directory maps: Map-backed cells in the staff directory handle icon loading more safely, including guarding against duplicate icon registration when lists re-render.
• Client bootstrap: The app entry path reduces unnecessary telemetry from the map SDK during normal use and keeps icon loading predictable on staff views.
• Build and dependencies: Tooling and dependency updates support the above fixes; an automated audit-fix workflow helps keep security patches on a predictable cadence.

Bug Fixes

• Staff pages: Resolved iterable and map-icon errors that could break staff directory rows when skills or map data were in unexpected shapes.

What's New

• Layer names and precincts: Map layers use plain-language names and short tooltips; precinct-related presentation is refactored so boundaries and labels stay understandable at a glance.
• Deeper zoom: Maximum zoom on the live map increases in two steps so teams can move from neighbourhood context down to street-level detail when reviewing incidents and places.
• Crime hex and viewport: Crime density respects the visible map area more consistently, and your last-used layer choices carry sensibly between sessions so you are not resetting filters every visit.
• Map and safety alignment: The live map shares behaviour and client-side caching with safety routing, including precinct shading where enabled, so the two experiences feel like one product rather than two maps.
• National crime layer: A national crime view helps compare local activity to wider patterns where your organisation turns it on; related reporting flows are shaped with privacy expectations in mind (including how report detail is surfaced).
• Live map and billing UI: In-product billing prompts on the live map are removed so the map stays focused on operations; subscription context instead lives with your billing and access tools.
• Access control — person-related fields: Optional person-related columns can appear in the investigation table and in PDF exports when your policy allows; additional structured fields can be included on PDF exports where configured.
• Billing and special access: Copy and gating around subscriptions, paid capabilities, and special access is clearer for admins and finance.
• Survey programme attribution: Surveys can show attribution where the GIG Quest programme applies.
• Payment integration: A third-party subscription checkout integration was merged, then reverted the same day so production returns to the previous, simpler billing story—avoiding a half-enabled payment path in the app.

Improvements

• Merge train: Multiple feature and staging merges land the live-map enhancement branch together with access-control export work, billing UX, and programme integrations so environments do not sit half-upgraded.
• Loading order: Background work is sequenced so the most important map layers become usable sooner during busy sessions.
• Precinct boundaries: Boundary visibility and reuse of cached boundary data are faster and more predictable when zooming or returning to an area.

Bug Fixes

• Precinct display: Fixed cases where precinct boundaries did not appear as expected until the map finished loading or until you changed zoom.

What's New

• Expose tripPurpose and notes in paginated export and export flows
• Enrich person information fields in paginated response
• Add person information fields to access control export skill

Improvements

• Release integration: stripe-integration, feature/export-flow-access-info, feature/access-control-person-info.

Bug Fixes

• Sync Stripe customer profile from organisation data

What's New

• Risk layer labels: Weight labels on risk-related map layers now match the current scoring architecture so what you read on the map matches the model behind it.

Improvements

• Release integration: Several staging merges bundle site-validation work, scoring tweaks, map-loading progress, and access-control–related fixes so customers receive a coherent upgrade rather than piecemeal patches.
• Local map data loading: While map-related data is prepared on the device, you see explicit progress instead of a frozen screen; controls for places and standard map overlays stay hidden until the underlying data is ready to use.
• Layer toggles: Place and overlay toggles are easier to notice and use, including when only part of the stack has finished loading.
• Organisation context for routing: Safety routing loads organisation profile data more reliably so route and overlay behaviour matches your tenant configuration.
• Safety routing complexity: The routing screen is simplified to reduce confusing states and edge-case interactions during live use.

Bug Fixes

• Safety routing: Addressed crashes in mixed map views when toggling overlays or switching modes quickly.
• Neutral intent scoring: When no specific intent is selected, overlays use a consistent, uniform treatment; modifier signs and weights for place-related overlays are corrected for scoring consistency.
• Site and hex data: Validation and streaming behaviour for site editing and large hex payloads are more reliable—fewer silent failures when saving sites or when data arrives in chunks.

Improvements

• Community context: Incident reports can be loaded at startup and merged into the hex graph for richer analysis.
• Scoring weights: Report, POI, and area context weights are tunable and wired consistently across scoring modes.
• Configuration: External configuration can drive scoring behaviour without code changes where supported.
• Engine modes: Optional second-stage engine can be enabled selectively; startup is staged to reduce peak memory use on smaller instances.
• Cache layout: Memory-efficient cache structure for large hex datasets.

Bug Fixes

• Scoring: Corrected weight application and modifier sign handling in edge cases.
• Quality: Test and validation fixes for routing and scoring pipelines.

What's New

• Add access control visit info fields to licence disc exports

Improvements

• Release integration: fix/scoring-v2, feature/ingestion-error-handling, feature/export-flow-access-info.
• Add count-collections script (hex-grid + reports); use top-level await
• Harden ingestion script error handling

What's New

• Billing workspace polish: Morning follow-up copy and layout refinements make subscription, paid features, and special-access messaging easier to scan for admins who live in this screen during renewals and audits.

Improvements

• Site editing validation: Site create/edit flows treat optional fields as optional in the schema, with clearer validation when required data is missing—reducing saves that silently drop data or fail late in the flow.
• Hex payload streaming: Large hex responses stream with a faster parser (less per-character overhead) and a JSON parser tuned for chunked payloads so the UI stays responsive while data arrives.
• Safety routing layout: Routing and site editing share a simplified routing chrome and tighter Edit site validation so common tasks stay on one screen without nested confusion.
• Staging and fix trains: Multiple staging merges land site-validation and hex-streaming fixes in coordinated batches so related behaviour ships together.

Bug Fixes

• Hex streaming and map readiness: Streaming fixes ensure large hex datasets can load incrementally; guards prevent calls into the map before layers exist, avoiding crashes when toggling views quickly. One change was briefly reverted and then re-applied so the final behaviour matches what QA signed off.
• Map layers: Edge cases where code touched map layers before initialisation completed are handled safely across safety and live-map flows.

Improvements

• Browser access: CORS and HTTPS settings aligned so web apps can call the service reliably in production.
• Large payloads: Timeouts tuned for very large aggregate responses so clients don’t fail mid-request.
• Memory: Safer default limits on constrained hosts to reduce out-of-memory restarts during heavy workloads.
• Routing: Wider initial search corridor and clearer road-only behaviour; optional toggles documented for operators.
• Operations: Deploy pipeline improvements with health checks and rollback paths when a release misbehaves.

What's New

• Extend billing summary and shareable checkout flow

Improvements

• Release integration: fix/scoring-v2.

Bug Fixes

• Hex scripts, weather util, runbook updates

What's New

• Billing workspace: Administrators get a dedicated area to review subscription status, paid-capability messaging, and who holds special access—so finance, security, and operations share the same picture without hunting through settings.
• Scoring and routing integration: The scoring follow-up tightens the unified crime-hex client, adjusts the safety-routing hex detail drawer, and updates routing and precinct clients so hex scoring and precinct loading stay aligned after the March scoring push.

Improvements

• Staging merges: Staging integrations land the scoring fixes alongside the billing workspace so upgrades receive both in one pass.

Bug Fixes

• Safety routing authentication: Removes brittle non-null assumptions and tightens auth handling so legitimate sessions are not rejected during routing transitions.

What's New

• Signed-in hosting: Deployment configuration aligned with standard platform authentication for protected environments.
• Cache bootstrap: Optional automatic download of hex cache data on startup for faster cold starts.
• Private artefacts: Support for authenticated download of cache packages from private release storage when configured.

Improvements

• Resilience: Service continues with sensible cache fallbacks when the primary data store is temporarily unavailable.
• Health checks: Single consolidated health registration for monitors and load balancers.

What's New

• Add stripe billing backend foundation

Improvements

• Release integration: fix/scoring-v2.

Bug Fixes

• Stabilize billing checkout draft creation

What's New

• Hex resolution and scoring: Crime hex cells use consistent resolution rules and scoring alignment end-to-end so colours, modifiers, and labels match when you compare cells or switch between map and analysis views.
• Unified crime hex integration: The shared crime-hex pipeline is wired through resolution, scoring, and display so analysts are not fighting two different interpretations of the same cell.
• Neutral precinct overlay: You can choose a None overlay mode for precinct choropleth shading—boundaries remain visible without a risk tint when you need a clean basemap for screenshots or briefings.
• Radar structure: Group radar extracts a dedicated main content surface and data hook so feeds and metrics stay readable during live review and future iterations are safer.

Improvements

• Report detail architecture: Report detail, underlying data fetch, and comments move behind focused hooks so the page scales as fields and commentary grow without fragile prop drilling.
• Precinct loading and readiness: Precinct choropleth waits for the map to be ready before painting, with automatic boundary enablement when the view is stable—fewer races where shading appears on a half-initialised map.
• Precinct client configuration: Precinct-related clients accept clearer options and overrides for routing-side loading so environments can point at the right services without one-off forks.
• Routing-side precinct fetch: Precinct boundaries for safety routing can be loaded through the same routing-oriented path as the rest of the safety map, reducing duplicate fetches and mismatched timing.
• Data and export types: Supporting types for analytics visualisation, skill reports, and CSV export paths are normalised so downstream exports stay consistent when schemas evolve.
• Onboarding and shell styling: First-step organisation basics and shared styling tokens are updated so early setup flows stay visually aligned with the rest of the enterprise shell.

Bug Fixes

• Scoring v2 display: Scoring consumes the latest server-side scoring payload, applies proportional weighting for place-related overlays, and fixes labels so patrol and analysis teams see what they expect.
• Widgets and embedded maps: Mini maps, bar-chart widgets, template dialogs, intelligence graphs, and global component search behave more reliably when data is sparse or when layouts reflow on smaller screens.
• Precinct local cache: The on-device precinct cache bumps its schema version and clears stale routing-side cache entries on upgrade so you are not stuck with outdated boundaries after a release.

What's New

• Precinct support: Cached precinct boundaries and related endpoints for location-aware routing and scoring.
• Hex scoring detail: Finer-grained hex scores with points-of-interest and open-map context where available.
• Scoring v2: Clearer curve and risk presentation; proportional use of POI context; more consistent segment scoring.

Improvements

• Corridor limits: Returned limits now match the active route configuration.
• Risk display: Tier scaling aligned with the latest scoring model.

Bug Fixes

• Scoring display: Curve and velocity presentation improvements for analysis views.
• Documentation: Notes on how precinct-level crime distribution affects scores where relevant.

Improvements

• Release integration: fix/scoring-v2.
• Document precinct crime distribution limitation

Bug Fixes

• Scoring v2 — fractional crime counts, proportional POI in segment scorer

What's New

• Crime type context: Icons next to report titles in the sidebar make it easier to scan what each item refers to.

Improvements

• Release integration: precinct-drawer-map-ux.
• Project docs: README and onboarding copy now match how the app is built and deployed.

Bug Fixes

• Sensitive categories: Report types that should not appear in the public map or list views are filtered accordingly for privacy.

What's New

• Licence plate search: The access-control search experience for vehicles is redesigned for speed and clarity—fewer dead ends, clearer results, and behaviour tuned for high-volume monitoring shifts when plates stream in continuously.
• Resilient routing baselines: Safety routing evaluates baselines more safely when inputs are partial or timing is tight, with fallbacks when upstream health is degraded so the UI does not hard-fail mid-session.
• Session-friendly caching: Recently used routing and hex context is retained locally so returning to a tab or resuming after a short interruption feels like continuing—not starting over.

Improvements

• Staging integration: Multiple merges land the licence-search and routing-hardening work together so production picks up both behaviours in one upgrade path.

What's New

• Open mapping: Android uses an open map layer so map startup doesn’t depend on vendor keys; the map stack moves to MapLibre with community tile sources—no paid map API keys required for the default setup.
• Device support: Minimum Android version raised so the current JavaScript engine and maps run reliably on supported hardware.

Improvements

• Builds: Preview and production build settings aligned with hosted environment variables; CI runs Android and iOS with separate steps for clearer logs when something fails.
• API routing: Backend base URL is normalized for packaged builds so client calls reach the correct API surface.
• Branding: Home-screen icon padding adjusted so the mark fits cleanly on iOS and Android launchers.

Bug Fixes

• Maps: Restored a stable map provider path on Android where the previous choice could crash on startup.
• GPS & forms: Location permission and accuracy failures surface clearer messages when creating sites and routes.

What's New

• Hex corridors: Routes can follow corridor geometry aligned to the hex grid for clearer analysis.
• Road snapping: Paths respect road geometry where configured.
• Validation: Stricter checks before routes are returned to callers.

Improvements

• Documentation refreshed for teams connecting routing, scoring, and validation workflows.

Improvements

• Release integration: feat/access-control-licence-search.

Bug Fixes

• Resolve licence plate search returning no results
• Type openmeteo API response for strict build

What's New

• Safety hex table and routing: Navigate hex-based safety analysis in a dedicated table view with resolution and scoring utilities that match the rest of the crime-hex model.
• Hex detail drawer: Open a full detail drawer for a cell so analysts see scoring, modifiers, and context in one place instead of splitting attention across panels.
• Hex and simulation caches: Dedicated stores keep hex crime data, simulation state, and scoring profiles available across steps so repeat lookups and what-if checks stay fast.
• Live-map places streaming: Community and open map-sourced places load through a streamed, parallel fetch so dots appear as each source returns instead of blocking the whole map.
• Map filters and safety path finder: Map filter state moves forward a generation alongside sidebar, dialog, and sheet updates so filters, layers, and the safety path finder stay coherent.

Improvements

• Unified crime hex client: Shared helpers for listing cells, pulling detail, and expressing safety intent keep the live map, safety flows, and hex tooling aligned.

What's New

• Wolf Patrol: First drop of the mobile app for field patrols—sites, routes, and checkpoints in one place.
• Offline-ready: Queue and sync behaviour so spotty connectivity doesn’t block a shift.
• Maps & location: Map zoom and tighter GPS handling for clearer positioning on patrol.
• Ship pipeline: Automated builds for Android and iOS so releases stay repeatable.

Improvements

• Experience: Updated visuals—icons, motion, and layout—for quicker reading outdoors and in vehicles.

What's New

• Safety Router: New service repository for routing and safety analysis workflows used by the platform.

What's New

• Device auth, crime hex tiles/detail, POI overpass, patrol socket and scripts

Improvements

• Release integration: feat/device-auth-crime-hex-poi-patrol.

What's New

• Safety routing workspace: Plan and review routes in a focused area with scoring context and simulation-style movement feedback where your organisation enables it, backed by a dedicated client for route and corridor insight.
• Report discovery and detail: Report lists, filter menus, and detail views are rebuilt so moving between analytics summaries and individual incidents feels like one flow—with saved filter utilities and tests to keep behaviour stable as data grows.
• Live map and reports sidebar: The live map and reports sidebar stay aligned with what you have selected in time and scope, including crime-hex context where configured.
• Shared navigation and chrome: Sidebar routes, time sliders, loading screens, and multi-select controls are brought in line so routing, reports, and map sessions share the same interaction patterns.

Improvements

• Map and time controls: Map pages respect report selection and time sliders more consistently during long monitoring sessions, with fewer mismatches between the map and the report list.

Bug Fixes

• Reports analytics: Analytics panels stay in sync with the rest of reporting; loading states follow the active theme; filters remain robust when underlying data refreshes or when you switch tabs quickly.

What's New

• Add safety routing, report enhancements, and crime hex improvements

Improvements

• Release integration: feature/performance-optimizations.
• Align report counts and improve report flow

Improvements

• Responsive directory: User tables adapt to tablet and phone widths with sensible column priorities and horizontal scroll where needed, so directory and role work is usable away from a large monitor.
• Invite experience: The invite modal presents fields and validation feedback more clearly, reducing back-and-forth when onboarding administrators or operators.
• Search: User search responds more predictably as you type and when combining filters, so large directories stay navigable during busy onboarding periods.
• Empty and edge states: When no users match filters—or when the directory is genuinely empty—copy and actions explain what to do next instead of leaving a blank grid.
• Release integration: Staging and performance-optimisation work is merged into the main line so environments pick up the latest platform fixes alongside these admin improvements.

Improvements

• Release integration: feature/performance-optimizations.
• Mark AdminUserService constructor-injected deps as readonly

What's New

• Map markers, precinct API, config updates
• Mobile layout, report loading, safe areas, accessibility
• Precinct boundaries/stations fixes and persistence

Improvements

• Release integration: precinct-drawer-map-ux.
• Precinct drawer, map UX improvements, transparent overlay
• Update marker icons and Map component

Bug Fixes

• Precinct stats cache, toggle sync, FAB positioning

What's New

• Streaming boundaries: Large precinct boundary sets load progressively with clear progress so the map stays usable instead of blocking on one big fetch.

Improvements

• Safety intent: Switching safety intent on the map refreshes scores in one coordinated pass so the overlay and legend stay aligned.
• Zoomed-out views: When you are far out on the map, precinct context loads with optimisations tuned for wide viewports.

What's New

• Response compression and police precincts cache
• Warm precinct cache at startup and cache /all-scores
• Allow CORS origins via env and add compression types
• Precincts streaming and no full-doc cache to avoid OOM

Improvements

• Release integration: feature/performance-optimizations.

Bug Fixes

• Move @types/compression to dependencies for production builds
• Remove eager precinct cache warming to prevent OOM on Render
• Safety score logic for zero crime and edge cases

What's New

• Precincts and safety paths: Teams can work with police precinct context and safety-path planning in a dedicated module aligned with your site and map setup.
• Report editing: Reports support incident dates and a visible edit trail so changes are auditable without digging through raw logs.
• Vehicle monitoring: Radar-style views use clearer bucketing and metric cards, with more reliable flag tracking during investigations.

Improvements

• Access control loading: Analytics and heavy tables load in stages with caching so returning to a tab or re-running a view feels faster under load.
• Export dialogs: Export limits are tighter where large extracts could strain operations, with clearer descriptions so users know what to expect.

Bug Fixes

• Export limits: Large export requests are constrained more predictably, with messaging that matches the limit behaviour.

What's New

• Police precincts, SAPS crime stats, analytics scale plan, and pathfinding module
• Propagate dateRange to all pipelines, parallelise queries, add indexes
• Flagged vehicle stats, data quality fixes, snapshot versioning, parallelised compute
• Add incident date tracking and report editing with audit trail

Improvements

• Release integration: feature/performance-optimizations.

Bug Fixes

• Restore pagination limit to 10000 for exports, wrap parse in try/catch
• Handle overflow in passengers $toInt when field contains non-numeric data

Improvements

• Release integration: feature/performance-optimizations.
• Reduce startup log noise
• Add README from actual codebase (stack, scripts, main areas, env)

Improvements

• Patrol: Event lists and management views load and filter more smoothly; export modals support broader patrol reporting needs, including PDF-oriented flows where your organisation uses them.
• Crime hex: The hex detail experience on the live map is refined for clarity and responsiveness during long review sessions.
• Access control exports: Export dialogs and downstream PDF generation are tuned for consistency with patrol and investigation exports.

What's New

• Optimize queries, add indexes, and enrich patrol events

What's New

• Hex and heat visualisation: The live map can show crime-related activity in hex cells with heat-style emphasis so teams can spot concentrations at a glance.
• Dedicated exploration: A focused hex-grid experience lets analysts adjust colours and controls when reviewing patterns outside the main map chrome.
• Detail drawer: Opening a cell surfaces more context in one place instead of jumping between disconnected panels.

Improvements

• Layer behaviour: Map filters and stores stay aligned when hex views are toggled with your other live-map layers.

What's New

• Add H3 hexagon grid and CW unified crime hex modules

Improvements

• Release integration: feature/performance-optimizations.

Improvements

• Release integration: fix/live-map-undefined-reports-filter.

Bug Fixes

• Live map filters: Avoids a crash when filters run before all report rows are available.

What's New

• Areas and checkpoints: Site areas and checkpoint markers are easier to place and adjust on the map, with sidebars that stay readable while you work across large sites.
• Site creation and editing: New and enhanced site flows use clearer dialogs and layout so long forms are less error-prone during setup.

Improvements

• Map interaction: Checkpoint markers and area outlines behave more predictably when panning and zooming during site planning.

What's New

• Improve reference checker, add geocode module and site/area validation

Improvements

• Release integration: feature/performance-optimizations.

Improvements

• Patrol routes: Route calculations and map interactions are optimised so planning and review sessions stay responsive on large sites.
• Broader performance: A focused optimisation sweep reduces unnecessary rendering and state churn across modules that saw heavy use after the February expansion.

What's New

• Optimize status counts and remove obsolete scripts
• Add route/schedule validation and block deletions when in use

Improvements

• Release integration: feature/performance-optimizations.
• Fix SonarLint issues (complexity, duplicate imports)

What's New

• Access control intelligence: Investigations gain map- and timeline-based views with playback-style controls and clearer analytics for vehicles, people, and patterns—without the clutter of older debug-style panels.
• Community reports: Report analytics and tables are rebuilt around saved filters, reusable components, and faster paths to charts and timelines.
• Executive and admin dashboards: The executive summary and related admin views are reorganised for steadier loading and clearer sections across community and enterprise snapshots.
• Agent groups: Nested routes and refreshed feeds, radar, insights, skills, and output screens make large group workspaces easier to navigate.
• Patrol: Patrol scheduling and status views gain dedicated building blocks for assignments, routes, and schedules so operations teams see status at a glance.
• Hotspot radius: Adjust hotspot coverage between five and five hundred metres on the map, with patrol table alignment fixes so columns stay readable.

Improvements

• Performance pass: Rendering and data handling are tuned across feeds, patrol, and schedules so heavy pages stay usable during peak use.

What's New

• Mega update for frontend parity

Improvements

• Release integration: feature/performance-optimizations.

Bug Fixes

• Fix prod build and add POI copy script

What's New

• Map filtering and markers: The live map uses a dedicated filter model with clearer marker treatment so layers and incidents stay legible at city scale.
• Report list cache: Recent report lists reuse cached data where it helps, cutting repeat waits when you move between map and list views.

Improvements

• Site management: Creating and editing sites is more consistent across new and enhanced flows, with validation and layout that behave better on long forms.
• Group exports: The export experience for agent-group output is streamlined for common CSV workflows.

What's New

• Forensic overhaul of live map reports API, POI resilience, and skill-report group export

Improvements

• 1 merge to main (including staging integration).

What's New

• Integrated release: The live-map places-of-interest work and the enterprise device workspace are merged into the same production path, so teams upgrading pick up both without juggling separate branches.
• Alert triage board: Device and alert boards emphasise recent activity and map context so operators can see what needs attention first.
• Delivery visibility: Alert-flow logs surface when messages fail to reach recipients, making it easier to spot configuration or routing issues early.

Improvements

• Device workspace: Locations, alerts, people-on-device, and live activity behave consistently with the rest of the February map and monitoring updates.

Bug Fixes

• Multi-recipient delivery: Addressed cases where alert flows needed more reliable handling when multiple recipients are involved.

What's New

• Enrich alert payloads and reduce stream noise

Improvements

• Release integration: feature/alarm-kanban-staff-alerts, fix/alert-flows-multi-recipient-delivery, feature/poi.
• Merge main into feature/poi - combine POI and alarm collections

Bug Fixes

• Harden recipient delivery and event routing
• Soft-delete alert flows and harden matching

What's New

• POI layer: Points of interest appear on the live map with filters you can tune by source and category so teams can focus on the facilities and places that matter for an operation.
• Detail drawer: Selecting a place opens a structured detail view with the key facts in one place instead of hunting across screens.
• Responsive loading: Map updates debounce sensibly when the view moves, and list data is cached by viewport so revisiting an area feels quicker.

Improvements

• Report context: Related report detail can surface faster from cache when you already viewed the area, keeping investigations flowing on busy shifts.

What's New

• Add POI layer, community green dots, OSM blue dots, filters

What's New

• Restore rich-map intelligence and SLA APIs

Improvements

• Release integration: feature/alarm-ingest-step1.
• Revert "feat(report): restore rich-map intelligence and SLA APIs"

What's New

• Device locations and alerts: See where hardware sits on a map, track alert activity in one place, and manage who is associated with each device when your organisation uses the device workspace.
• Installer linking: Pairing or linking installers favours a phone-first flow so field staff can complete steps on mobile without fighting desktop-only paths.
• Access control columns: Choose which columns appear in licence-plate investigations and keep labels consistent so teams share the same view during live monitoring.

Improvements

• Filters and plate views: Filter work runs in parallel where it helps, slow calls time out more gracefully, and plate rows stay in sync when data streams in during busy shifts.
• Device workspace: Live updates keep device, location, and alert views current while you work across tabs.

What's New

• Add intelligence batch endpoint for licence disc logs
• Add phone linking endpoint with onboarding gate

Improvements

• Release integration: fix/access-control-backend-scale.
• Merge remote-tracking branch 'origin/main' into feature/alarm-ingest-step1
• Align emergency coverage service formatting

Bug Fixes

• Remove text index hint and clean up repository

What's New

• Patrol and areas: Site areas and map views use a clearer colour system and smoother map behaviour; patrol scheduling steps are easier to work through when building routes and timings.
• Access control at scale: Filtering and plate-based views load more predictably under heavy use, with caching so returning to a tab does not feel like a cold start.

Improvements

• Live map and mini map: Map widgets respond more reliably when panning and zooming around sites and patrol context.
• Incident report links: Admin-facing link management stays aligned with the latest public incident form behaviour.

Bug Fixes

• Public incident form: Resolved issues that could affect step state and incident classification on the public submission flow after recent layout changes.
• Access control stability: Addressed edge cases in filter loading, timeouts, and row updates so long sessions stay usable during peak monitoring.

What's New

• Add alerts and locations realtime
• Optimize pipelines and observability; fix(area): delete semantics and types
• Add user linking and alert fanout flows
• Include provider context in alert fanout

Improvements

• Release integration: fix/access-control-backend-scale.
• Access control: runbook, structured logging, socket emit logging

Bug Fixes

• Scale and observability for backend

Improvements

• Release integration: fix/emergency-coverage-always-recompute.

Bug Fixes

• Always recompute coverage overlap for decisioning

Improvements

• Reporting and monitoring: Upgrades through this release pick up the full public incident experience—map-based location, broader categories, and admin shortcuts—alongside the February feed and radar performance work, so operations teams get both in one deployment.

Improvements

• Release integration: feature/perf-optimisation-plan.

Improvements

• Group message feeds: Long-running feeds stay responsive when many messages are on screen—lists load in manageable batches and avoid unnecessary redraws so operators can keep scrolling without the interface stalling.
• Radar monitoring: The radar view updates less aggressively when nothing material has changed, so the picture stays readable during busy periods.
• Licence plate views: Plate-based lookups in access control benefit from the same performance pass, so repeated checks feel quicker during investigations.

What's New

• Optimise radar analytics and intelligence search

What's New

• Public links: Support for incident report URLs that can be opened outside WhatsApp when sharing with partners or the community is appropriate.

What's New

• Device workspace: Enterprise teams can open a central view for alarm and device hardware—see what is online, review recent signals, and drill into a device without leaving the main platform navigation.
• Alert-focused logs: The default experience centres on alert history with filters you can adjust in place, so operators spend less time hunting for the right slice of activity.

Improvements

• Live activity: Device and alert views refresh as new information arrives, keeping floor teams aligned with what is happening on the ground.
• Device actions: Common follow-ups are easier to reach from the log and device detail flow, reducing friction when acknowledging or triaging issues.

What's New

• Add alarm ingest and device registry
• Manage devices and events
• Add realtime sockets and editable router IDs

What's New

• Map-based location: Community members can place and refine where an incident occurred using an interactive map in the public flow, with clearer copy for tap-to-place and selected location.
• Expanded incident categories: The public form aligns with a broader set of incident types—covering safety, property, infrastructure, and service-related issues—so reports match how organisations classify work on the ground.
• Admin sidebar: Incident report management is linked from the main admin navigation for quicker access alongside other enterprise tools.

Improvements

• Group messaging views: Feeds, chats, insights, exports, radar, and skills screens for messaging groups are more consistent and easier to scan during live monitoring sessions.

Improvements

• Release integration: feature/public-incident-report-links.
• Merge remote-tracking branch 'origin/staging'
• Add auth context tracing for agent groups

Bug Fixes

• Preserved platform roles set by authMiddleware so organisationOwnershipMiddleware no longer overwrites them with an empty array, which was starving AccessLevelService and causing 401s on Group Agents.
• Propagate auth context to agent group scope
• Fallback to organisation roles for platform access
• Widen org id type in platform roles
• Resolve auth context in org ownership

What's New

• Guided detail prompts: After someone describes an incident in their own words, the form can suggest additional details they might add—presented in plain language with a short loading state so expectations stay clear across supported languages.
• Streamlined steps: The public incident flow drops a redundant incident-type screen so reporters move from location and timing to details and contact with fewer taps.
• Date and time picker: A dedicated date-time control makes it easier to record when the incident happened without awkward manual entry.

Improvements

• Dashboard: Report widgets are hidden by default on the main dashboard so the first view stays focused; noisy diagnostic logging is reduced for a calmer experience in normal use.

Bug Fixes

• Group messaging exports: Fixed naming and collection wiring for exported group lists used in messaging workflows so the right data appears in exports and access checks stay aligned with your configuration.

What's New

• Add LLM categorisation

Improvements

• Release integration: fix/auth-orgid-staff-fallback.

Bug Fixes

• Derive org id from staff fallback
• Reuse platform roles from auth context
• Inherit org roles for users

What's New

• Report LLM: Execute endpoint for the report summarization and analysis path, consistent with how other LLM features are exposed.

What's New

• Map pin workflow: Community members using a public incident link can place and adjust a location pin as part of the flow so reports line up with the right place on the map.
• Emergency staff messaging: The emergency requests experience supports richer templates and clearer handling when messages need to go to your team in urgent situations.
• Access control skills: New options in staff setup support access-control workflows that need person-related logging where your organisation enables them.
• Site skill assignment: Only organisation-wide administrators can assign certain skills at the site level, reducing the risk of misconfiguration for large deployments.

Improvements

• Admin navigation: Incident report tools are easier to reach from the main admin sidebar so teams can jump straight into link management after the initial rollout.
• Public report reliability: The public submission flow is more resilient when the page updates or reloads, so users are less likely to lose context mid-report.
• Web submissions: Browser-based submissions are attributed correctly as web traffic for clearer reporting and follow-up.

Bug Fixes

• Public incident flow: Addressed cases where hooks behind the public form could get out of sync, which could interrupt or confuse the submission experience.
• Device attribution: Fixed how web visitors are labelled so analytics and downstream views stay consistent for public reports.

What's New

• Add log_person_info skill to agent state mapping
• Add parent-first pagination
• Use geo index for area matches
• Add allowlisted stream logging
• Support public incident report links
• Add stream log allowlist

Improvements

• Release integration: feat/person-only-access-control-flow-skill.

Bug Fixes

• Allow optional geo index metadata
• Accept device string

What's New

• Public incident report form: Shareable links let community members submit incidents online with location and language support.
• Admin link management: Create, manage, and track report links in the enterprise platform.

Bug Fixes

• Themed scrollbars: Scrollbars use theme variables so they stay consistent in light and dark mode.

Improvements

• 2 merges to main (including staging integration).

Bug Fixes

• Increase pagination limit to 10000 for exports

Improvements

• Message synchronization: Messages sent from the platform are now properly synchronized across all systems, ensuring conversation continuity.
• Message ordering: Improved timestamp handling ensures messages appear in the correct order throughout emergency conversations.

Bug Fixes

• Duplicate messages: Fixed issue where location sharing created multiple duplicate messages in conversation history.
• Message context: Resolved issue where platform messages weren't visible to automated systems, ensuring proper conversation context is maintained.

What's New

• Emergency workflows: Platform messaging and agent controls support emergency scenarios with clearer handoff.
• Message path: Backend support keeps third-party message flows aligned with platform expectations.
• Connector email: Outbound connector mail uses plain text for broader client compatibility.

Improvements

• Message ordering: Emergency conversations list messages in a stable time order for accurate reading.
• Message tracking: Conversation history stays in sync so teams see a single coherent thread.
• Release integration: coverage-fixes-dedup.

Bug Fixes

• Duplicates & sequencing: Fewer duplicate location entries and better ordering in fast-moving emergency threads.
• History sync: Message history reflects platform activity flags consistently.

What's New

• Persist org coverage and emergency records

Improvements

• 1 merge to main (including staging integration).

Bug Fixes

• Inject organisation id for area repo

Improvements

• Default report view: Reports show the full set by default instead of automatically narrowing to organisation areas unless you choose that filter.
• Verified filter: Verified status only filters when you explicitly enable it.
• Connector testing: Connector test flows include users linked through staff records so test sends match real access patterns.
• Release integration: general-fixes.

Bug Fixes

• Report filtering: Resolved cases where area filtering hid reports when users expected the full list.
• Sort behaviour: More consistent ordering across report views and live updates.

What's New

• Email connectors: Configure automated emails from report events with organisation-specific templates and a safe way to send test messages before going live.
• Email templates: Placeholders pull in organisation details so outbound mail stays on-brand.
• Report intelligence: Explore report relationships with graphs and maps; see alert counts alongside intelligence views.
• Patrol comments: Add notes to patrol events so teams keep clearer context for follow-up.

Improvements

• Event flow: More reliable processing and delivery for connector-triggered emails.
• Report context: Intelligence views surface richer relationships between reports and alerts.
• Formatting: Tidier report and route presentation across related views.
• Release integration: feature/connectors, feats/report-intel-and-fixes.

Bug Fixes

• Connectors: Tighter filtering so emails only go to the right people for the organisation; staff-linked users are included where expected.
• Reports: Reports list behaviour respects “show all” vs area filters; verified filters only apply when you turn them on.
• Real-time updates: Socket payloads carry a sortable timestamp so live ordering stays consistent.

What's New

• Add backend schema and listener
• Enforce report type enums
• Add organisation email placeholders

Improvements

• Add intelligence debug diagnostics

What's New

• Access Control Intelligence: Explore vehicle activity with a visual timeline and map trail for faster investigations.
• Event details from timeline: Open full event details directly from timeline entries for deeper context.

Improvements

• Faster investigation flow: Jump from timeline entries to detailed event views without leaving the intelligence screen.

Bug Fixes

• Map trail stability: The intelligence map now loads reliably when search results appear.

What's New

• Include disc log id in track points

Improvements

• Release integration: feature/ac-intelligence-graph-map.
• Add access control intelligence tags

Bug Fixes

• Resolve complete info via discLogId fallback

What's New

• Add agent group connection telemetry for uptime tracking
• Add vehicle intelligence graph endpoint

Improvements

• Release integration: feature/agent-group-connection-telemetry.

Bug Fixes

• Lazy-init telemetry service to avoid DB access at module load

Improvements

• Release train: Staging work is merged to main so production tracks the same feature set as other Wolf services.

What's New

• Insights execution: Server-side endpoint to trigger LLM-based insight generation with the same guardrails as the rest of the API.

Bug Fixes

• Broken shared links: Some shared links could land on a 404 instead of the intended page; routing now resolves correctly.
• Link parameters preserved: When a redirect is needed, query parameters are kept so users stay in the right context.

Improvements

• Resilient link handling: The app tolerates unusual or malformed link formats from external sources.

What's New

• Step-by-step survey flow: Surveys now guide users through a clearer, sequential experience with progress indicators to reduce drop-off and confusion

Improvements

• Staff directory feedback: Creating or editing staff now shows clearer validation feedback to help resolve common data issues quickly

Bug Fixes

• Patrol management stability: Fixed a crash that could occur in patrol management when some details were missing or incomplete

Improvements

• Clearer confirmations: Users now get more consistent acknowledgement when completing key actions, helping reduce confusion and repeat attempts
• More consistent staff handling: Improved staff-related behaviour to reduce duplicates and improve day-to-day reliability
• Security hardening: Updated dependencies to address a reported security issue

Bug Fixes

• Report interactions: Fixed an issue where a “More Info” action could reference the wrong report in some cases

Improvements

• Surveys: Smoother flows with less friction in common paths.
• Staff records: More consistent handling so profiles stay aligned across the product.

Bug Fixes

• Patrol views: Fewer missing details or inconsistent staff metadata when viewing patrols.

What's New

• Survey Outcomes Viewer: Browse all survey submissions in a detailed table showing dates, ratings, concerns, environmental factors, and locations
• Analytics Dashboard: Visualise survey data with KPIs, rating distribution charts, and breakdowns of top concerns and environmental factors
• Interactive Map: View survey submission locations on a map with two visualisation modes:
  • Heatmap View: See density of submissions and areas of concern at a glance
  • Pins View: Explore individual submissions with clustered markers for better performance

Improvements

• Unified Filter Panel: All views (Outcomes, Analytics, Map) now share the same filter controls for a consistent experience
• Advanced Filtering: Filter surveys by:
  • Specific public link
  • Submission type (public or staff)
  • Date range
  • Day and night safety ratings
  • Specific concerns (e.g., drug activity, loitering)
  • Environmental factors (e.g., poor lighting, graffiti)
• Multi-field Search: Search across notes, concerns, and environmental factors simultaneously
• Submission Details: Click any submission to view full details including notes and precise location
• Pagination: Efficiently browse large numbers of submissions with cursor-based pagination

User Experience

• Organised tabbed interface separating Links, Outcomes, Analytics, and Map views
• Colour-coded ratings help quickly identify areas of concern
• Collapsible advanced filters keep the interface clean while providing powerful filtering options

What's New

• Survey outcomes: Paginated submissions with strong filtering options.
• Analytics: Aggregates for ratings, concerns, and environmental factors.
• Geography: Locations for heatmaps and pin maps.

Improvements

• Filtering: By public link, submission type, date range, safety thresholds, concerns, environmental factors, and more.
• Search: Search notes, concerns, and environmental fields together.
• Performance: Indexes and cursor-based pagination for large datasets.

Improvements

• Release integration: Branches for the Render-native media pipeline are merged into main in a single coordinated drop.
• Code quality: Type-safety fixes so builds stay strict as the media surface grows.

Improvements

• Real-time feed updates: Message feeds now update automatically as new messages arrive, reducing manual refresh while monitoring
• More responsive monitoring: Improved performance and reliability when viewing active feeds over longer sessions

Improvements

• Real-time updates: Feeds update as messages arrive with less manual refreshing.
• Session stability: Better behaviour during longer monitoring sessions.

Improvements

• Transcription: Temporary audio handling can target a directory you control (for disk quotas, permissions, or ephemeral storage).

What's New

• Improved Image Loading: Access control event images now load more reliably with a dedicated loading state
• Enhanced Security: Images are now served through secure, time-limited URLs for better protection

Improvements

• Better Image Gallery: Image preview dialogs now show a loading indicator while images are being prepared
• Consistent Experience: Both vehicle licence disc images and person information images now use the same secure loading mechanism

What's New

• Custom Brand Styling: Organisations can now configure their own brand colours, logo, font, and corner roundness for public-facing surveys
• Organisation Profile Page: New dedicated page for managing organisation details and public branding settings
• Font Selection: Choose from 11 professional fonts including Inter, Roboto, Poppins, and more
• Dynamic Corner Styling: Select from five border radius options (None, Small, Medium, Large, Full) that apply across all survey elements

Improvements

• Streamlined Settings: Renamed "Profile" to "Organisation Profile" in the sidebar for clarity
• Light Mode Defaults: Public surveys now default to a clean light theme matching the platform styling
• Reset Functionality: "Reset to Wolf Defaults" now preserves your enabled/disabled setting while resetting styling values
• Consistent Styling: Border radius settings now apply uniformly to all cards, buttons, and chips on surveys

How It Works

1. Navigate to Settings → Organisation Profile
2. Switch to the Public Branding tab
3. Toggle Custom Branding on to enable your settings
4. Customise colours, logo, font, and corner roundness
5. Click Save Branding to apply changes to your public surveys

What's New

• Secure Image Storage: Licence disc and person information images submitted via WhatsApp are now stored with enhanced security
• Improved Reliability: Image uploads are tracked more reliably for better troubleshooting

Technical Details

Images submitted through access control flows now use secure storage references, ensuring consistent and reliable image delivery across the platform.

Bug Fixes

• Fixed report notification delivery: Resolved an issue where report notifications weren't being delivered to some users due to template configuration mismatches
• Improved chat reliability: Fixed an issue where follow-up messages could fail after sending images, ensuring conversations continue smoothly
• Corrected template language handling: Fixed template delivery failures caused by incorrect language code formatting

Improvements

• Separate enterprise and community templates: Report notifications now use dedicated templates for Enterprise and Community users, ensuring consistent delivery across both platforms
• Better error messaging: Improved error handling provides clearer feedback when template issues occur

What's New

• Brand configuration: Organisations can save colours, logos, fonts, and styling; branding can be toggled independently of resets.
• Secure access-control images: Time-limited URLs for access-control imagery with cleaner storage and metadata.
• Persistent media in reports: Skill reports can carry durable media URLs so images and videos stay available over time.

Improvements

• Light-mode defaults: Default palette aligns with the platform light theme.
• Flexible reset: Resetting brand defaults keeps your enable/disable choice.
• Access-control migration: Existing images move to the new delivery model without breaking older records.
• Report media reliability: Attachments use redundant, long-term accessible links where applicable.

What's New

• Durable attachments: Media linked to skill reports is kept in a form that survives retries and replays.
• Storage efficiency: Message history layout adjusted to reduce redundant data for high-volume threads.

What's New

• Click Analytics: See how many times each public survey link has been clicked
• Geographic Breakdown: View which countries and cities your survey clicks are coming from
• Real-time Refresh: Manually refresh stats with a single click

Improvements

• Mobile Submit Button Fix: Fixed issue where submit button could be hidden on mobile devices with notched screens
• Safe Area Support: Added proper viewport handling for iPhone X and newer devices

Technical Details

Analytics are powered by Short.io's tracking — no additional setup required for links that have a Short.io URL configured.

What's New

• Link analytics: Total clicks and last-click time per public survey link.
• Geographic insights: Country and city breakdowns for link traffic.
• Short links: Analytics work for short-URL configurations.

What's New

• Unified media path: Images and files are accepted and processed by the application service directly instead of relying on a separate upload worker.

Improvements

• Queues and throughput: Tweaks to background handling so bursts of media don’t stall conversations.

What's New

• Safety Survey Settings: New admin page for creating and managing public survey links
• Named Survey Links: Create multiple links with descriptive names to track different survey campaigns
• One-Click Sharing: Copy survey URLs or short links directly from the dashboard
• Public Survey Page: Beautiful mobile-optimized survey accessible to anyone with the link — no login required
• Staff Survey Access: Quick access button to open the authenticated staff survey from the settings page

Improvements

• Mission-Driven Messaging: Survey now explains the purpose — "We're on a mission to make the world safer"
• Cleaner Interface: GPS status only appears when action is needed, keeping the focus on the survey
• Multi-language Public Surveys: Public surveys support all four languages (English, Afrikaans, isiZulu, isiXhosa)
• Sidebar Access: Safety Survey settings now accessible from the Business Tools menu

When to Use Each Survey

Staff Survey (requires login):

• Internal team doing structured neighborhood audits
• When you need to know who submitted each response
• Training staff on the survey process
• Higher accountability requirements

Public Links (no login):

• Crowdsourcing safety data from the community
• Sharing on social media, flyers, or QR codes
• Campaign-specific tracking with named links
• Maximum participation with minimal friction

What's New

• Safety Survey Tool: A new mobile-optimized survey for collecting community safety perception data while walking through neighborhoods
• Day & Night Ratings: Rate how safe an area feels during the day versus at night, with intuitive emoji-based rating scales
• Multi-language Support: Switch between English, Afrikaans, isiZulu, and isiXhosa with a single tap - perfect for South African community surveys
• Observed Concerns: Quickly tag issues noticed in an area like loitering, poor lighting, or abandoned buildings
• Automatic Location Capture: GPS coordinates are captured with each submission for mapping and analysis

Improvements

• Apple Weather-style Design: Beautiful day and night themed cards make it immediately clear which question you're answering
• Rapid Data Collection: 5-second cooldown between submissions allows quick surveying while walking
• Offline-Ready Language: Language preference is saved locally so it persists between sessions
• Compact Interface: All controls fit on a single mobile screen for efficient field use

What's New

• Survey data collection: Capture geolocated safety perception responses with structured concerns and language tracking.
• Public survey links: Share anonymous survey links without login; create multiple named links per organisation for campaigns.
• Short URLs: Friendly short links for social and print.
• Anonymous attribution: Responses tie back to your organisation while respondents stay anonymous.

Improvements

• Permissions: Submissions respect organisation role rules.
• Link management: Enable, disable, or rotate links; rate limiting and risk-based verification reduce abuse.
• Standalone module: Survey capability remains self-contained for easier operation and evolution.

Improvements

• Agent prompts: Conversation templates updated alongside the merged prompt-improvements work.
• Media integrity: Skill reports no longer retain empty media references when an upload does not complete successfully.
• Metadata: File size and related fields are normalized consistently for downstream processing.
• Optional backfill: Maintenance tooling added for reconciling older media records where needed.

Bug Fixes

• Upload edge cases: Hardening around partial or failed media transfers.

What's New

• Message Feeds: Monitor messages from multiple groups side-by-side in a multi-column layout
• Feed Decks: Save and switch between named feed configurations to organise different monitoring views
• Flexible columns: Add groups as columns (including multi-select) to build a monitoring view quickly

Improvements

• Live updates: Feeds refresh automatically as new messages arrive, reducing the need to manually refresh while monitoring
• Layout polish: Improved wide-layout containment to keep the feed board easier to use on smaller screens

What's New

• Radar tab: A new analytics tool in Group Agents that helps you understand what's happening across your monitored WhatsApp groups
• Keyword tracking: Create a personal watchlist of keywords and phrases to monitor across your groups. See how often they appear and track changes over time
• Sentiment analysis: Get an automatic breakdown of message sentiment (positive, neutral, negative) to gauge the mood of conversations
• User activity insights: See which participants are most active and how their message patterns relate to your tracked keywords
• Trending terms: Discover the most frequently used terms across your monitored groups

Improvements

• Flexible time windows: Analyze messages over 7, 14, 30, or 60 day periods
• Feed-based scoping: Use your existing Feed decks to scope Radar analytics, or select individual groups
• Match type options: Choose between substring matching (finds "car" in "racecar") or exact word matching for more precise tracking
• Trend comparisons: See how keyword counts have changed compared to the previous period with delta indicators

What's New

• Message Feeds: A new way to monitor multiple WhatsApp groups simultaneously with a customizable multi-column layout
• Feed Decks: Create and manage multiple named feed configurations to organize different monitoring views
• Persistent Media: Images and media in message feeds now load reliably from secure cloud storage

Improvements

• Bulk Message Loading: Messages across multiple groups load efficiently in a single request
• Per-User Configurations: Feed layouts are saved per user, so each team member can customize their own view

What's New

• Real-time patrol tracking: Patrol events now update instantly across all connected users, providing live visibility into patrol activities
• Live access control updates: Access control changes are now reflected in real-time across the system

Improvements

• Better staff creation: Improved validation when creating staff members across different sites
• Faster access control: Optimized performance for browsing and filtering access control records

Bug Fixes

• Fixed an issue where site validation could fail when creating new staff members

What's New

• Daily Predictions: Get expected crime counts for any specific day, not just weekly aggregates

• Hourly Breakdowns: Understand when crime is most likely throughout the day with peak hour analysis

• Detailed Crime Types: See breakdown within categories (e.g., within "Theft", what percentage is vehicle break-ins vs shoplifting)

• Confidence Intervals: Statistical bounds on predictions (e.g., "expect 12 crimes, 95% confidence between 8-17")

Two Complementary Models

The API now provides insights from two models in a single response:

New Response Fields

Predictions now include:

• Expected daily counts by crime bucket
• Peak hours for each crime type with risk multipliers
• Top 15 specific crime types within each category
• Time factors (day-of-week effect, seasonal patterns, recent trends)
• Historical comparisons (vs day-of-week average, monthly average, recent 4-week average)

Use Cases

• "What's the worst hour for theft in this area?" - Peak hour analysis
• "Is today above or below normal?" - Historical comparison
• "What specific crimes should I worry about?" - Detailed type breakdown
• "How confident are you in this prediction?" - 95% confidence intervals

Performance

All 10 cities now have baseline models that train in under 60 seconds, enabling rapid updates as new data becomes available.

What's New

• Expected Crime Counts: The model now predicts not just "what type of crime" but also "how many crimes" to expect in each area

• Per-Type Volume Breakdown: Get expected counts broken down by crime category (e.g., "expect 3 thefts, 2 assaults this week")

Before vs After

Before: "This area has 35% theft probability, 25% assault" After: "This area has 35% theft (expect ~3 incidents), 25% assault (expect ~2 incidents), 8.5 total crimes expected"

All Cities Updated

Every city model has been retrained with dual prediction capabilities:

Improvements

• 8 cities now under 5% error (up from 5)
• Boston improved from 7.85% to 5.82% error
• SF improved from 6.26% to 4.79% error
• Average error across all cities: 4.74%

API Response Updates

All prediction endpoints now include:

• expected_total: Total crimes expected in the time period
• expected_counts: Breakdown by crime type
• count_mae: Model accuracy for volume predictions

What's New

Three new cities complete our initial 10-city goal:

• Boston: 80 grid cells covering greater Boston area
• Toronto: 238 grid cells - first Canadian city!
• Denver: 263 grid cells across the metro area (achieved #5 ranking!)

International Expansion

The platform now covers 3 countries:

• United States (8 cities)
• United Kingdom (London)
• Canada (Toronto)

Full Coverage Summary

Platform Statistics

• 10 cities across 3 countries
• 2,766 grid cells total coverage
• 7 million+ crime records processed
• 5 cities under 5% prediction error
• 46 API tests all passing

What's New

Four new US cities added to the prediction platform:

• San Francisco: 30 grid cells covering the SF peninsula
• Los Angeles: 585 grid cells - our largest coverage area
• Seattle: 93 grid cells across the metro area
• Philadelphia: 163 grid cells covering the city

Coverage Summary

Total Coverage

• 7 cities across 2 continents
• 2,185 grid cells total
• 6.1 million crime records processed
• 4 cities under 5% prediction error

Improvements

• Standardized crime categories work consistently across all US cities
• Fixed handling of rare crime categories in cities like Seattle
• Los Angeles model handles largest grid (585 cells) efficiently

What's New

• Real-time report updates: Reports now refresh automatically when new data arrives, keeping dashboards up-to-date without manual refresh

Bug Fixes

• Fixed display issues in spatial reports
• Improved report stability for production environments

What's New

• New York City Coverage: Full crime prediction support across Manhattan, Brooklyn, Queens, The Bronx, and Staten Island

• Best-Performing Model: NYC achieves our lowest prediction error yet at 3.73%, thanks to extensive historical data

Coverage Summary

NYC Highlights

• 446 hexagonal grid cells covering all boroughs
• 104 weeks of prediction data available
• Sub-4% error across most crime categories
• Automatic routing works seamlessly with existing integrations

Three Cities, One API

The unified endpoint now serves predictions for three major cities across two continents. Existing integrations continue to work unchanged - just send coordinates and receive predictions.

What's New

• Redesigned dashboard system: A completely rebuilt dashboard that loads faster and provides more reliable data snapshots
• Enhanced data exports: Better collection support and example values when setting up data exports

What's New

• London Coverage: Full crime prediction support for Greater London using Metropolitan Police data

• Unified API with Auto-Routing: Send coordinates from any supported city - the API automatically detects location and routes to the correct model

• International Expansion: First cross-Atlantic deployment, proving the system works with different police data formats and crime categorizations

Coverage Details

How It Works

No changes needed to your integration. Simply send coordinates:

• Chicago coordinates receive Chicago predictions
• London coordinates receive London predictions
• Unsupported locations return a clear "not covered" response

London-Specific Features

• Adapted to UK crime categorization from Met Police data
• 590 prediction cells covering Greater London
• Full probability distributions matching Chicago format

What's New

• Hexagonal Grid System: Migrated from custom square grids to the industry-standard H3 hexagonal system used by major rideshare and logistics companies

• Globally Unique Cell IDs: Every prediction cell now has a unique identifier that works anywhere on Earth, enabling multi-city expansion

Improvements

• Prediction error reduced to 4.32% - 4% improvement from the previous 4.50%
• Training time reduced by 33% - Faster model updates and iterations
• Cleaner spatial relationships - Each cell has exactly 6 neighbors at consistent distances

Why Hexagons?

Unlike squares where corners are 40% farther from the center than edges, hexagons have uniform neighbor distances. This creates more consistent spatial patterns for the model to learn, resulting in better predictions.

What This Enables

• Consistent cell coverage across any city globally
• Standardized cell sizes (~5 km² per cell)
• Foundation for multi-city expansion
• Industry-compatible cell identifiers for integrations

What's New

• Single Location Predictions: Send any latitude/longitude in Chicago and receive a complete crime probability distribution with risk level assessment

• Batch Predictions: Query up to 100 locations in a single request with summary statistics across all locations

• Route Risk Assessment: Calculate cumulative crime risk along a travel route with segment-by-segment breakdown and customizable risk weighting

New Endpoints

What You Get Back

For each prediction request, you receive:

• Probability breakdown across 6 crime categories
• Dominant crime type and confidence level
• Overall risk classification (Low, Moderate, High, Very High)
• Grid cell information for the queried location

Use Cases

• Delivery routing: Identify safer routes for drivers
• Real estate: Compare neighborhood safety profiles
• Urban planning: Understand crime distribution patterns
• Personal safety: Plan safer travel routes

Model Performance

• 4.50% mean absolute error (new best!)
• All crime categories balanced below 6% individual error
• Strong correlation between predicted and actual patterns

Improvements

• More robust data exports: Export flows are now more reliable with better error handling
• Improved data visualization: Fixed collection handling and added missing filter operators
• Code cleanup: Removed unused roster collections for a cleaner system

What's New

• Redesigned Crime Categories: Restructured crime classifications into more balanced and meaningful groups for better predictions

• Probability Distribution Focus: Shifted focus from simple classification to full probability distributions - now providing the complete picture of crime likelihood

Improvements

• Overall prediction error reduced to 4.74% - When we say "35% chance of theft," the actual value is typically between 30-40%
• Improved prediction quality across all crime types
• Better balance between common and rare crime categories

New Crime Categories

• Theft & Burglary: All theft-related incidents grouped together
• Battery & Assault: Physical altercations and violence
• Vehicle & Property Damage: Auto theft, vandalism, and property crimes
• Serious Violent: High-severity incidents requiring immediate attention
• Other: Remaining categories including drug offenses and public order

What This Means For Users

Previously, the system would tell you "this area is high risk for property crime." Now it tells you the complete breakdown: "35% theft risk, 25% assault risk, 20% quiet, 15% vehicle damage, 5% other." This full distribution enables better decision-making.

What's New

• More Accurate Probability Predictions: The model now outputs well-calibrated probability distributions, meaning when we say "35% chance of property crime," you can trust that number

• Smarter Handling of Low-Activity Areas: Added intelligent classification for areas with minimal crime activity, preventing false positives in quiet neighborhoods

• Honest Geographic Boundaries: Refined the prediction coverage to focus only on populated areas, ensuring accuracy metrics reflect real-world performance

Improvements

• Prediction accuracy improved to 82% on validated land areas
• Probability calibration error reduced by 80%+ (from 0.55 to 0.09)
• Distribution quality score improved significantly (Brier score: 0.72 to 0.15)

Technical Highlights

• Switched from daily to weekly prediction windows for more stable patterns
• Implemented soft probability labeling for smoother predictions across activity levels

What's New

• Full Patrol Management: Complete system for scheduling, tracking, and managing security patrols
• Dynamic Checkpoint Tracking: Checkpoint intervals now calculate based on actual distance rather than fixed times
• Automated Notifications: WhatsApp notifications for patrol start, progress updates, and abandonment alerts
• Manager Overrides: Managers can now override patrol schedules and settings as needed

Improvements

• Smart Scheduling: Patrol schedules now anchor to your configured start-of-day time
• Guard Reminders: Automatic nudges for guards who pause mid-route
• Abandoned Patrol Detection: System now flags patrols that never started or were abandoned

Bug Fixes

• Fixed issue with inconsistent data formatting
• Resolved warning message display issues
• Improved location accuracy on warning events

What's New

• Saved report filters: Save your frequently used report filter configurations and quickly apply them later
• Multi-area statistics: View aggregated statistics across multiple areas in a single report

Improvements

• Better export organization: Improved code structure for a smoother export experience

Improvements

• More reliable authentication: Improved login reliability with better fallback handling
• Simplified staff access: Removed unnecessary role restrictions for a smoother experience

What's New

• Patrol assignment overrides: Override scheduled patrol assignments for specific situations without changing the underlying schedule
• Improved schedule management: Better handling of schedule conflicts and changes

What's New

• Dynamic staff status: View real-time status of staff agents directly in the system
• Improved patrol management: Better validation for patrol routes and overrides

Bug Fixes

• Fixed issues with alert flow handling
• Improved reliability of export flows across organizations

What's New

• Patrol events search: Search through patrol events with filtering by status and full pagination support
• Enhanced patrol events: Improved patrol event tracking with better reporting capabilities

Improvements

• Better multi-backend support: Improved handling when running multiple server instances
• Cleaner data model: Removed legacy fields for a more streamlined experience

What's New

• Soft delete for staff: Staff members can now be deactivated instead of permanently deleted, preserving historical records

Improvements

• Better organization handling: Improved analytics reliability when switching between organizations
• Rostering updates: Better alignment with scheduling requirements
• License number normalization: Consistent formatting for license numbers

Bug Fixes

• Fixed various access control issues
• Improved schedule permission handling

What's New

• Patrol system v1: Complete patrol management system for organizing and tracking patrol activities
• Rostering foundation: Initial support for staff rostering and scheduling

Bug Fixes

• Fixed patrol route editing permissions
• Various bug fixes and stability improvements

Bug Fixes

• Fixed permission issues with gate keeper functionality
• Added support for both British and American spelling conventions

Improvements

• Various stability and bug fixes

What's New

• Site area integration: Link patrol events and activities to specific areas within sites
• Enhanced patrol events: Improved tracking and management of patrol events

Bug Fixes

• Fixed permission handling inconsistencies
• Resolved issues with permission field configurations

What's New

• Patrol schedules: Create and manage patrol schedules at the organization or site level
• Improved organization access: Better handling of user organization membership

Improvements

• Simplified patrol schedule creation workflow
• Better support for organization-wide schedules
• Consistent field naming conventions across the platform

Improvements

• Gatekeeper optimization: Significant performance improvements to the gatekeeper system
• Better authentication handling: Resolved conflicts between different authentication routes

Bug Fixes

• Fixed authentication issues affecting certain routes
• Improved patrol authentication to match other route patterns

What's New

• Patrol routes: Define and manage patrol routes for security staff
• Staff assignment to patrols: Assign staff members to specific patrol routes
• Large payload support: Handle larger data transfers for complex operations

Improvements

• Better data visualization: Improved date range handling and group ID support in visualizations

What's New

• Patrol routes management: Create, edit, and manage patrol routes for security operations
• Patrol schedules: Set up recurring patrol schedules with customizable timing
• Staff assignment: Assign staff members to patrol routes and schedules

What's New

• Data visualization date ranges: Filter visualizations by custom date ranges for better insights
• Collection name support: Consistent handling of collection names across visualizations
• Group filtering: Filter data by groups for more focused analysis

Improvements

• Centralized connections: Improved reliability with centralized data source management
• General stability and performance improvements

What's New

• Remarketing campaigns: Create and manage remarketing campaigns to re-engage your audience
• Message templates: Design reusable templates for consistent communication
• Audience building: Build audience segments based on user behavior and attributes
• Opt-out analysis: Track and analyze opt-out patterns to improve engagement

Improvements

• Phone number identification for better user tracking

What's New

• Access Control System: Complete access control workflow for Enterprise tier with visitor management and vehicle verification
• Voice Agent: Hands-free incident reporting and panic alerts via voice calls
• Faster AI Responses: AI agents now process multiple actions simultaneously for quicker responses
• System Health Monitoring: Real-time monitoring ensures the system is always available
• Shorter Links: Cleaner message formatting with shortened URLs

Improvements

• Smarter Message Routing: System now intelligently routes between conversational and action-based interactions
• Automatic Failover: If one AI provider is unavailable, the system automatically switches to a backup
• Better Address Formatting: More consistent and accurate address display

Bug Fixes

• Fixed chat history appearing twice in some cases
• Fixed handling of stickers and reactions in messages
• Resolved video message processing issues

What's New

• GateKeeper system: Complete access control system for managing entry points, visitors, and permissions
• Column filtering: Filter and customize which columns appear in your access control views
• Analytics dashboard: View access control statistics and insights

Bug Fixes

• Fixed permission handling for administrator roles

Improvements

• Dynamic alert flow updates: Alert flows now automatically reload when their configuration changes
• Better date handling: Improved date string conversion in report filters for more accurate filtering
• Referenced fields in emails: Alert flow emails can now include related data from referenced fields

What's New

• Report comments: Add comments and notes to reports for team collaboration and context

Bug Fixes

• Fixed issues with reports not loading correctly
• Improved area count display in reports

What's New

• Permission management: Define granular permissions for different actions and resources
• Role-based access control: Create roles with specific permission sets and assign them to users
• Access level control: Services now respect user access levels throughout the platform

What's New

• Webhook support: Trigger external systems via webhooks when alert conditions are met
• Send on any change: Alert flows can now trigger on any data change, not just specific conditions
• License disc naming: Updated license plate terminology to license disc for clarity