CAOS Care

CAOS Care is the care-focused vertical of the CAOS ecosystem.

It is intended to support senior-care and assisted-living workflows through resident-centered memory, staff support, care-plan visibility, safety escalation, documentation, a resident room node (local CAOSCare computer behind the TV) with voice/audio and room hardware, staff clients (tablet / phone / computer) running the appropriate workspace, predictive behavior-change awareness, and a practical CCE-lite trust layer. Resident rooms are not tablet-based — see docs/CAOSCARE_PRODUCT_BASELINE.md.

Start here for agents

Before making claims or changes, read in order:

1. AGENTS.md
2. docs/CAOSCARE_PRODUCT_BASELINE.md — canonical durable product truth
3. docs/PROJECT_STATE.md
4. docs/REPO_MAP.md
5. docs/BUILD_STATUS.md / docs/CURRENT_NODE_STATUS.md
6. docs/CAOS_CARE_AGENT_ONBOARDING_CONTRACT.md (hardware section is HISTORICAL / SUPERSEDED), docs/CCE_LITE_TRUST_LAYER_PROPOSAL.md, and task-specific contracts

The full boot sequence is in AGENTS.md and docs/CAOSCARE_PRODUCT_BASELINE.md §9.

CCE-lite trust layer

CAOS Care should use the CAOS Council Engine direction as a practical care-safe profile, not as a heavy multi-model council on every resident/staff interaction.

The first useful version is:

CCE-lite = intent classifier + risk gate + verifier + receipt + human escalation

This means CAOS Care should prioritize:

• residents being heard faster;
• staff being routed clearer;
• family/staff communication being verified before it matters;
• sensitive outputs staying inside human-supervised boundaries;
• leadership having receipts for what happened.

Full council mode belongs later in admin, incident review, policy, research, architecture, and complex decision-support workflows.

See docs/CCE_LITE_TRUST_LAYER_PROPOSAL.md.

Core doctrine

CAOS Care is not a replacement for clinicians, caregivers, licensed medical professionals, emergency services, or regulated clinical judgment.

CAOS Care is a governed assistive platform for:

• care-plan visibility
• resident reminders
• staff empowerment
• resident safety alerts
• family/staff communication support
• documentation and receipts
• behavior-change detection as advisory signal
• the resident room node + staff-client model (docs/CAOSCARE_PRODUCT_BASELINE.md §2/§3); wearable ingest
• escalation support under human oversight
• CCE-lite routing, verification, and receipt-backed trust workflows

Required stance

• Human care remains primary.
• AI output is advisory unless explicitly approved by authorized humans.
• Safety-critical workflows require receipts, escalation paths, and confirmation gates.
• Private resident data must be protected.
• No feature is complete without acceptance criteria and regression notes.

Current state

This section is a snapshot from early onboarding and understates where the project is. For real current state read docs/PROJECT_STATE.md (recent dated entries) and docs/CURRENT_NODE_STATUS.md. The repository is an active multi-surface CAOSCare codebase with a deployed backend/frontend and a live resident-room bring-up host — not a blank build surface.

Public website crawl was attempted from ChatGPT tooling, but no crawlable CAOS Care site content was available at the time this README was updated. Future agents must inspect the deployed website directly when accessible and update the repo map/onboarding contract with verified page content.