| name | os-04-public-safety |
| description | Operating-system orchestrator skill for **Public Safety, Justice Operations, and Emergency Response** (national operating system #4). Use this skill whenever work touches this sector's mission — Prevent harm, respond to emergencies, maintain order, and recover from acute incidents — to understand the jobs to be done, the human/AI/robot division of labor, the accountable human boundaries, and which specialized role skills to deploy. Trigger this even when the user names a specific task in the domain rather than the sector itself. |
Operating System 04 — Public Safety, Justice Operations, and Emergency Response
Layer: National operating system (#4 of 23) · Personnel model: human-owned, AI- and robot-augmented
Cross-references: 00-framework/ (shared concepts, teaming pattern, accountability), source map Country-Economy Core Jobs To Be Done.md
Mission
Prevent harm, respond to emergencies, maintain order, and recover from acute incidents.
When to use this skill
Load this skill when a task concerns public safety, justice operations, and emergency response. It gives an agent the sector map: the outcomes that must be produced, who owns them, what can be automated, and where human accountability is non-negotiable. From here, route to the specific role skills under roles/ for execution.
Core Jobs To Be Done
These are the durable outcomes this operating system must reliably produce, written as trigger → response:
- When someone is in danger, receive the signal, dispatch help, and stabilize the situation.
- When crime occurs, investigate, preserve evidence, and support prosecution or restorative processes.
- When fires, floods, earthquakes, pandemics, or industrial accidents occur, coordinate multi-agency response.
- When infrastructure fails, prioritize rescue, shelter, utilities, medicine, and public communication.
- When risk can be reduced, inspect, educate, enforce, and prepare.
The universal lifecycle, applied
Every job in this sector moves through the same seven steps. Use it as a checklist when designing or executing work here:
- Sense reality — gather data, observe conditions, inspect sources, listen to people.
- Interpret reality — diagnose, forecast, model risk, prioritize.
- Decide — choose policy, design, action, allocation, escalation, or tradeoff.
- Mobilize — assign labor, budget, materials, rights, permissions, logistics, schedule.
- Execute — perform the work in digital or physical space.
- Verify — test, audit, measure, inspect, certify, and learn.
- Govern — maintain legitimacy, safety, accountability, continuity, and trust.
Human role families (who owns the work)
- 911 dispatcher, emergency manager, incident commander.
- Firefighter, EMT, paramedic, search-and-rescue specialist.
- Police officer, detective, crime analyst, forensic technician.
- Probation officer, corrections officer, victim advocate.
- Safety inspector, fire marshal, disaster recovery specialist.
These remain human-owned. AI personnel and robots augment them; they do not replace the accountable owner.
Labor-market grounding (how these roles are advertised)
The human roles this operating system staffs appear on job boards with concrete, checkable signals. The AI-personnel and robot skills here are designed to support these advertised roles, not to replace the accountable human in them.
- Advertised titles & seniority ladder: Recruit/officer/EMT → detective/paramedic/senior → sergeant/lieutenant/captain → chief; dispatcher → comms supervisor; emergency-management coordinator → director.
- Skills, tools & tech employers list: CAD (computer-aided dispatch), RMS (records management), NIMS/ICS, body-cam/evidence systems, NCIC, GIS.
- Qualifications, certifications & licenses: POST certification (police), state EMT/Paramedic (NREMT), Firefighter I/II, EMD, FEMA ICS/NIMS, CEM (certified emergency manager).
- KPIs / metrics in postings: Response and call-answer times, case clearance rate, incident outcomes, mutual-aid readiness, safety.
- Where these roles are posted: GovernmentJobs, National Testing Network/PoliceApp, USAJOBS, local agency sites, Snagajob (some support roles).
Grounding reflects 2026 job-posting conventions across LinkedIn, Indeed, Dice, ZipRecruiter, Glassdoor, USAJOBS, GovernmentJobs, and specialized boards, spot-verified against public listings and O*NET/BLS. Re-verify specifics — especially pay, certifications, and licenses — against live postings before operational use.
AI personnel in this operating system (deployable role skills)
Each of the following has a dedicated, extensive skill under roles/. Deploy them under the named human supervisor:
- Emergency call triage assistant — classifies incoming calls, extracts location and severity, and supports dispatch. (supervised by 911 dispatch supervisor; skill:
roles/emergency-call-triage-assistant/)
- Dispatch optimizer — allocates and routes responders against live demand. (supervised by emergency manager; skill:
roles/dispatch-optimizer/)
- Incident summarization agent — maintains a live common operating picture and after-action logs. (supervised by incident commander; skill:
roles/incident-summarization-agent/)
- Crime pattern analyst — detects spatial-temporal crime patterns and links cases. (supervised by crime analyst; skill:
roles/crime-pattern-analyst/)
- Evidence chain-of-custody assistant — tracks evidence handling and flags integrity gaps. (supervised by forensic technician; skill:
roles/evidence-chain-of-custody-assistant/)
- Forensic media review agent — reviews video/audio/digital media for relevant events. (supervised by detective; skill:
roles/forensic-media-review-agent/)
- Disaster scenario planner — models hazard scenarios and resource needs. (supervised by emergency planner; skill:
roles/disaster-scenario-planner/)
- Public alert drafting agent — drafts multilingual, accessible public warnings. (supervised by public information officer; skill:
roles/public-alert-drafting-agent/)
- Resource allocation agent — matches shelters, supplies, and crews to needs. (supervised by logistics chief; skill:
roles/resource-allocation-agent/)
Work-system completeness (the work around the core work)
The core roles above are necessary but not sufficient. For each material JTBD, check which ancillary services are required:
| Family | Required support question | Reusable catalog |
|---|
| Enable | Do practitioners have the evidence, knowledge, data, tools, access, and skills they need? | _catalogs/enabling-work/ |
| Integrate | Who owns dependencies, handoffs, queues, decision preparation, and stakeholder alignment? | _catalogs/enabling-work/ |
| Assure | What needs independent quality review, challenge, testing, risk, safety, legal, or audit work? | _catalogs/enabling-work/ |
| Adapt | How are alternatives generated and operational experience converted into improvement? | _catalogs/enabling-work/ |
| Sustain | Who maintains administration, capacity, wellbeing, coverage, assets, and institutional memory? | _catalogs/enabling-work/ |
Do not clone every support role into this sector. Choose embedded, shared, platform, federated, or temporary support according to demand, specialization, consequence, and context. Every ancillary service must name the core JTBD and owner it serves, its trigger, deliverable, service level, decision boundary, outcome link, escalation, and retirement rule. See the Work-System Completeness Map.
Humanoid robot roles
- Hazardous entry, fireground supply movement, stretcher support, debris inspection.
- Shelter logistics, food/water distribution, sanitation support.
How these robots work (assumed architecture): each is an LLM-brained embodied agent — a multimodal LLM brain plans and issues physical actions as tool calls (e.g. grasp, navigate_to, place), executed by Vision-Language-Action policies trained on world models, robot gyms, and RLAIF. Fleets may share one brain model or mix specialized ones. A verified low-level safety layer can override unsafe actions independently of the brain. Full detail in 00-framework/ and _catalogs/humanoid-robots/.
Non-humanoid autonomous machines
Self-driving vehicles, equipment, and drones for this sector (LLM-planned; physical actions as tool calls; ODD + teleoperation fallback):
- Search & response drone — search for people, map incidents, and deliver overhead situational awareness in emergencies. (autonomous machine skill:
autonomous/search-response-drone/)
How these machines work (assumed architecture): each is a non-humanoid autonomous machine — a foundation/LLM planning brain issues actions as tool calls (follow_route, dump_bucket, take_off, spray_zone, …) over a perception → prediction → planning → control stack trained on world models, driving/field simulation, and RLAIF. Each runs inside a defined Operational Design Domain (ODD) with a verified safe-stop and teleoperation fallback. Full detail in _catalogs/autonomous-machines/ and 00-framework/.
Human accountability boundary (must stay human-led)
Arrests, use of force, triage in scarce life-saving situations, sentencing, detention, and incident command remain human-led.
Treat this boundary as a hard constraint. Agents in this sector may sense, interpret, draft, model, monitor, and coordinate up to this line, then must hand off to an accountable human for the decision itself.
Division of labor (human / AI / robot)
- Human owner — accountable for goals, values, exceptions, relationships, signoff, and everything inside the accountability boundary above.
- AI personnel — research, draft, analyze, monitor, simulate, coordinate, document. Strongest on digital signals and repeatable decision support.
- Robot personnel — fetch, carry, inspect, clean, assemble, assist, enter hazardous spaces. Strongest on physical work in human-built environments.
- Control layer — permissions, audit logs, escalation thresholds, incident reporting, evaluation.
- Public trust layer — explainability, appeal, privacy, bias testing, safety certification, labor-impact review.
Interfaces with other operating systems
This sector regularly depends on and feeds: Defense & Intelligence, Health & Care, Resilience & Continuity, Water & Sanitation. Coordinate handoffs explicitly; most systemic failures happen at the seams between operating systems.
Strategic missions that draw on this sector
Beyond its own mandate, this operating system is composed by these cross-cutting strategic missions (the orthogonal mission axis — a mission pulls roles from several sectors toward one national objective):
Sector success metrics (illustrative)
- Coverage / reliability: the share of the population or demand reliably served.
- Quality / safety: defect, incident, and harm rates within tolerance.
- Cost / efficiency: unit cost and resource use trending down without eroding safety.
- Trust / legitimacy: public confidence, complaint resolution, and auditability.
- Resilience: time-to-detect and time-to-recover from shocks.
Failure modes to watch
- Monoculture / correlated failure — shared models or vendors failing in lockstep; require diversity and manual fallback.
- Cascading dependency — failures propagating from the systems listed above; map dependencies and design graceful degradation.
- Deskilling — losing the human bench that can run the sector manually; retain drills and manual modes.
- Agent-specific failure — fabrication, prompt injection, reward hacking, silent drift; keep the control layer independent.
- Speed mismatch — automated action outrunning human oversight; install circuit breakers for high-consequence steps.
Deskilling watch & keep-warm regime
Automating routine cases erodes three things over time: the human fallback bench (who runs this when automation fails), tacit / craft judgment (lost as the experienced cohort retires), and the learning ladder (juniors never get the cases they used to learn on). Job and role simulators are the primary countermeasure.
- Risk here: Dispatch and response depend on CAD; incident commanders lose improvisation under protocolized tools.
- Countermeasures: Manual-dispatch drills; full-scale exercises with technology disabled; sim-based skills currency.
- Role/job simulators (keep-warm): Incident-command and dispatch simulators; tech-down field exercises; EMS code-blue sims.
Dual-use simulators: the world models and simulation built to train the machines in this sector double as the keep-warm simulators that keep humans current and rebuild the learning ladder. Owned cross-sector by OS 22 (Resilience) and the _catalogs/simulation-training/ roles; the verified deterministic fallback in _catalogs/capability-optimization/ is its technical complement.
Adapting to any nation (context modifiers)
The jobs above are universal; how they are staffed is not. Public-safety roles act where the state's coercive power meets individual rights; error costs are asymmetric and borne by identifiable people. Bias and evidentiary standards are operational constraints, not compliance afterthoughts.
Re-read this sector through:
- Scale (city-state → federation): whether this role is unified or layered across local/regional/national tiers.
- State capacity (fragile → high-capacity): whether the owning institution exists and can be held to account, or the job is met by markets, households, NGOs, or donors.
- Income level (low → high): affordability of automation and the balance of subsistence vs. wage work.
- Formality (informal → formal): whether the people and assets this role acts on appear in any registry at all.
- Resource & geography: which hazards and dependencies dominate (water-scarce, flood-prone, landlocked, trade-dependent).
- Political system & legitimacy: where the human-accountability boundary actually binds and who may hold power to account.
How to operate in this sector
- Identify which Core JTBD the task serves.
- Select the role skill(s) under
roles/ that fit, and confirm the human supervisor.
- Run the work-system completeness check and add only the Enable, Integrate, Assure, Adapt, and Sustain services the core outcome requires.
- Run the lifecycle: sense → interpret → decide → mobilize → execute → verify → govern.
- Stop at the accountability boundary and route the decision to the accountable human.
- Log actions to the control layer and surface anything that trips a failure mode.
