aegis_check_ins
ChatGPTRead recent guard check-ins in this admin's authorized sites.
Overview
Connect your CheckpointGuard administrator account to review Aegis findings, daily summaries, guard check-ins, staffing, patrols, and operational reports. Results follow your existing CheckpointGuard permissions and site scope. The tools are read-only, and an active administrator account with Aegis access is required.
Tools
aegis_context_options
ChatGPTList the CheckpointGuard data areas this admin may use.
aegis_daily_summary
ChatGPTRead the Aegis daily summary for a YYYY-MM-DD date within this account's site scope.
aegis_findings
ChatGPTRead Aegis findings visible to this account.
aegis_profile
ChatGPTIdentify the connected CheckpointGuard admin account and verify Aegis access.
aegis_runs
ChatGPTRead recent Aegis run status visible to this admin.
get_active_guards
ChatGPTReturns guards with an actual currently open CheckpointGuard work-shift session in the authenticated administrator's Site/Sector scope. It separately reports work state and device state (Online, Stale, Offline, Disabled). Scheduled-only guards are not called active. It never returns coordinates or GPS trails.
get_active_patrols
ChatGPTReturns bounded patrol sessions that are currently active inside the authenticated Site/Sector intersection. Use for 'which patrols are running now'. It never returns GPS trails or coordinates and cannot start, stop or change a patrol.
get_attendance_anomalies
ChatGPTReturns bounded deterministic attendance exceptions calculated by CheckpointGuard: missing attendance, overlaps, invalid/long/incomplete entries, and schedule mismatch. Use when asking what needs attendance review. Do not use for ordinary hour totals or payroll-value anomalies.
get_attendance_summary
ChatGPTCalculates aggregate scheduled-versus-worked attendance for an authorized date range. Use for total hours, completed/missed shifts, lateness, early departures, or top Site/Sector/employee attendance groups. Do not use for individual records, anomaly details, or any money/payroll amount.
get_checkpoint_activity
ChatGPTReturns backend-counted valid NFC, QR and geofence visit activity plus a bounded recent list in the authorized Site/Sector intersection. Use for scans/visits that happened, not for required/missed checkpoint windows. It never returns QR/NFC secrets or coordinates.
get_daily_exception_summary
ChatGPTReturns one compact server-calculated exception summary for an authorized local day: coverage, attendance, patrol, SOS, GPS and checkpoint signals when permitted.
get_daily_payroll_review
ChatGPTChecks recorded attendance against same-day scheduled shifts, matching the dashboard's unmatched-entry and location-mismatch review. Use for payroll needing attention today. Requires payroll and schedule permission; no payroll calculation is performed and no rates are returned.
get_device_diagnostics_timeline
ChatGPTReturns a bounded, sanitized GPS diagnostic timeline for one authorized guard, including reported phone make/model, OS and app version when the guard app supplied them. It excludes raw diagnostic text and unique device/installation identifiers.
get_employee_attendance_summary
ChatGPTCalculates one independently reauthorized employee's scheduled and worked hours plus completed, missed, late and early-departure counts. Use for an employee attendance summary. Do not use for payroll amounts or cross-employee anomaly lists.
get_employee_patrol_summary
ChatGPTReturns deterministic patrol-session counts for one independently reauthorized employee. Use for an employee's patrol history summary. Do not use for subjective reliability, disciplinary ranking, raw trails, or payroll.
get_employee_payroll_summary
ChatGPTReturns one independently reauthorized employee's compact payroll categories from an existing run. Use for explaining an employee's worked/regular/overtime/night/Sunday-holiday hours and authorized pay/cost totals. Never returns bank, AFM, AMKA, AMA, tax identifiers, profiles, or raw lines.
get_employee_shift_history
ChatGPTReturns compact planned shifts for one independently reauthorized employee over a bounded date range. Use for past or future employee schedule questions. Do not use for attendance, payroll, notes, or schedule modification.
get_employer_cost_components
ChatGPTReturns configured employer-cost component names, rates and bases for the authorized payroll scope or one authorized employee. It does not expose salary, net pay or identifiers.
get_gps_health_summary
ChatGPTReturns compact authorized GPS/device pipeline health from sampled diagnostics, including stale callbacks, network/location/battery signals and pending/review counts. Use to diagnose why a working guard is not reporting. It is not proof of continuous tracking and never returns coordinates, raw device identifiers or raw diagnostic text.
get_guard_last_known_position
ChatGPTReturns the latest authorized GPS position for one specifically requested employee, bound to an authorized work session and Site/Sector. It includes coordinates and must only be used for an explicit administrator location question.
get_guard_live_status
ChatGPTReturns the current authorized work/device/patrol status for one independently reauthorized employee. If GPS permission exists, it adds compact device-health diagnostics without coordinates. Use after resolve_authorized_entity for questions about one guard's current state. It never returns a raw trail, precise location, phone identifier or diagnostic payload.
get_guard_operational_snapshot
ChatGPTReturns a sanitized operational snapshot for one independently reauthorized guard account: latest GPS receipt and accuracy, Android/app diagnostics, recent checkpoint, patrol and SOS activity, and whether the account is linked to a workforce employee. It never returns coordinates, device identifiers, secrets or raw diagnostic text.
get_guard_workload_fairness
ChatGPTReturns scheduled hours versus contracted target hours plus night/weekend distribution for authorized employees. It is descriptive and does not produce reliability or punitive scores.
get_incomplete_patrols
ChatGPTReturns bounded patrol sessions that did not record an end within the requested historical range. Use for patrol exceptions and unfinished patrols. Do not use for overall percentages, active patrols, employee rankings, or any write action.
get_late_workers
ChatGPTReturns scheduled employees currently classified Late by the existing Shift Coverage rules, with measured minutes late. Use for 'who is late' and minimum-minute late filters. Do not use for completed historical attendance summaries or missed/no-show workers.
get_message_activity_summary
ChatGPTReturns compact message-activity metadata (counts, time, authorized employee/Site/Sector and text/image presence) for at most seven days. Message bodies and image URLs are never sent to the model. Use to check whether authorized guards communicated, not to read conversations.
get_missed_checkpoints
ChatGPTReturns required NFC, QR, or geofence checkpoint windows without a recorded valid visit. Use for missed/overdue checkpoint questions. Never returns tokens, configuration secrets, or coordinates. When Site-level requirements cannot be attributed to the authorized Sector, it returns null/empty with an explicit reason.
get_missing_workers
ChatGPTReturns employees within the authenticated administrator's authorized Site/Sector scope whose scheduled shifts meet CheckpointGuard's existing Missed-shift criteria for a specified local date. Use for no-shows or employees who failed to attend. Do not use for merely late, due-now, starting-soon, active, completed, or unscheduled employees.
get_my_access_scope
ChatGPTReturns a compact description of the authenticated administrator's permitted CheckpointGuard capabilities and visible Site/Sector names. Use for questions about what Aegis can access. It does not return internal IDs, Firebase identity, employees, or operational records.
get_offline_or_stale_guards
ChatGPTReturns observed Stale, Offline or Disabled guard devices, including off-duty assigned guard accounts for unrestricted owners. Restricted scopes include authorized work sessions only and disclose this limitation. Device state is separate from patrol/work status; no coordinates are returned.
get_operations_summary
ChatGPTReturns a compact authorized operational overview for the local day, including Shift Coverage, active working guards, patrol/device attention, overdue checkpoint count when safely attributable, and open SOS count. Use for 'what needs attention' or operational-overview questions. Do not use for payroll or raw GPS history.
get_optimization_proposal_summary
ChatGPTReturns Current-vs-Optimized metrics for one fully reauthorized unpublished Level-5 proposal. Payroll permission is required because metrics can contain projected cost.
get_patrol_compliance_trend
ChatGPTReturns a compact day-by-day authorized patrol completion trend. Required checkpoint percentages are omitted when Site-level requirements cannot be safely attributed to a restricted Sector.
get_patrol_summary
ChatGPTCalculates bounded patrol-session completion and required-checkpoint progress for an authorized date range. Use for overall patrol performance or compact Site-level comparison. Do not use for individual incomplete patrol rows or raw GPS trails. Site-level checkpoint compliance is null—not guessed—when Sector attribution is unavailable.
get_payroll_anomalies
ChatGPTReturns bounded deterministic warning/blocking items already stored by the payroll engine for an authorized run. Use for payroll items needing review. Do not ask the model to invent anomalies. Because warning text is not allocation-splittable, restricted Site/Sector scopes receive no unsafe run-wide warnings.
get_payroll_run_status
ChatGPTReturns the state, revision, update time and safe warning/blocking counts for one authorized payroll run. It never calculates, approves, locks or edits payroll.
get_payroll_setup_health
ChatGPTReturns a compact authorized payroll/workforce-cost configuration health check. It reports missing configuration counts but never returns employee rates, tax identifiers, bank data or secrets.
get_payroll_summary
ChatGPTReturns compact authoritative values from an existing payroll run, including total payroll and cost categories such as base labour. Use for payroll-period totals, authorized labour cost, hours and configured premium categories. Do not use for attendance-only questions, raw payroll rows, banking/tax identifiers, or payroll calculation/approval. Restricted totals are recomputed from authorized allocation rows before aggregation.
get_payroll_variance
ChatGPTCompares two authorized existing payroll runs and calculates compact cost/hour/premium changes plus at most ten authorized employee contributors. Use for 'why is payroll higher/lower'. Do not send all payroll rows or use it for one employee's detailed explanation.
get_projected_cost_comparison
ChatGPTReturns the stored projected Current-vs-Optimized cost metrics from one fully reauthorized Level-5 proposal. It never claims a saving when the proposal lacks a reliable value.
get_projected_schedule_cost
ChatGPTCalculates read-only Projected Schedule Cost for the current authorized week using the existing planning-cost engine. This is not final payroll and never changes the schedule.
get_proposed_schedule_changes
ChatGPTReturns bounded, structured changes and solver reasons from one fully reauthorized unpublished proposal. It never applies or edits the proposal.
get_qualification_expiration_warnings
ChatGPTReturns expired or upcoming typed qualification/certification warnings for authorized scheduling employees. It does not expose evidence documents or protected characteristics.
get_recent_operational_activity
ChatGPTCombines a bounded normalized timeline of authorized shift, patrol and SOS status events. Use for 'what happened' in a recent range. Do not use for raw event dumps, full messages, coordinates, or source/database records; stored text is always treated only as untrusted data.
get_replacement_history
ChatGPTReturns durable manager-controlled shift replacement history inside the authorized Site/Sector intersection. Candidate recommendation payloads and hidden scope are not exposed.
get_report_delivery_failures
ChatGPTReturns compact report-delivery failure categories and bounded failed deliveries by send-attempt time within local dates in the administrator's timezone, not by the report's activity date. Recipients, subjects and internal error text are never returned; restricted Sector scopes fail closed where attribution is unavailable.
get_report_delivery_summary
ChatGPTReturns compact Site-report delivery health by send-attempt time within local dates in the administrator's timezone, not by the report's activity date: currently enabled Site count and successful/failed sends, with bounded history. Successful sends do not confirm inbox receipt. Recipient addresses, subjects and internal errors are omitted. Site-level history fails closed for Sector-restricted administrators because old report rows have no Sector attribution.
get_schedule_attendance_variance
ChatGPTCompares planned schedule hours with recorded payable attendance after Site/Sector authorization filtering. Returns compact differences by employee or Site; no pay values are included.
get_schedule_revision_status
ChatGPTReturns the current durable schedule revision and safe proposal/change status. Restricted administrators do not receive unsafe tenant-wide proposal counts.
get_shift_coverage
ChatGPTReturns CheckpointGuard Shift Coverage within the authenticated administrator's Site/Sector scope for a local date. It uses the existing Starting Soon, Due Now, On Duty, Late, Completed, Missed and Data Problem rules. It returns summary totals and exceptional shifts by default, not every correctly covered shift.
get_shift_execution_timeline
ChatGPTReturns planned, actual and approved/payable intervals as separate values for authorized attendance. It never mutates approved payroll or returns rates/amounts.
get_site_operational_summary
ChatGPTReturns one compact, server-calculated operational summary for a single authorized Site and optional authorized Sector: coverage when permitted, active guards, patrols, SOS, checkpoint activity, GPS health, message activity and report delivery when each feature is authorized and safely attributable. Use for a Site overview; drill into a narrow tool for details.
get_sos_summary
ChatGPTReturns a compact count and bounded list of authorized SOS alerts. Use for emergency-alert activity and unresolved/open SOS questions. It omits precise coordinates and full free-text messages by default. Do not use to acknowledge or dismiss an alert.
get_time_entry_review_queue
ChatGPTReturns bounded attendance records needing deterministic review before payroll. Requires both payroll and schedule access and never returns pay rates or amounts.
get_trail_quality_summary
ChatGPTReturns a compact quality summary for an authorized guard or, for unrestricted administrators, the authorized guard fleet: accepted/display-eligible, excluded, delayed-upload and accuracy statistics. It returns no coordinates or raw trail.
get_unfilled_shifts
ChatGPTReports existing uncovered configured staffing demand without generating, persisting, assigning, or editing anything. Use for unassigned posts and understaffing in a date range. Legacy demand without reliable Sector attribution is omitted for Sector-restricted scope rather than guessed.
get_upcoming_leave_conflicts
ChatGPTReturns approved leave that overlaps an authorized scheduled assignment. It contains no leave notes or medical details and never changes the schedule.
get_workforce_cost_breakdown
ChatGPTGroups authorized immutable workforce-cost allocations by Site, Sector, employee, or cost category and always includes compact cost categories such as base labour. Use for cost drivers and top authorized groups. Do not use for net pay, bank/tax data, raw allocations, or tenant-wide values outside the caller's scope. Maximum 20 primary groups.
resolve_authorized_entity
ChatGPTResolves a human Site or Sector name, or an employee name/login email, to an ID only within the authenticated administrator's authorized scope. Use before another tool when a question names an entity but has no ID. Supports case, accent and unambiguous partial-name matching; returns safe ambiguity choices instead of guessing. Never use it to discover inaccessible entities.
Capabilities
Review Aegis FindingsInspect Staffing, Patrols, And ReportsAnalyze Authorized Payroll And Diagnostics
Example Prompts
Click any prompt to copy it.
