Which integrations are available for this server?

Provides an Alexa+ MCP server for Echo Show that delivers aging-in-place care tools: cognitive check-ins with linguistic biomarker analysis, medication safety verification, ambient home-safety anomaly triage, and caregiver digest dispatch. It also integrates with Amazon Bedrock via the Bedrock Runtime Converse API to generate multi-agent care-circle summaries.

How do I use EchoCare Kinship?

1. Click on "Deploy Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@EchoCare Kinship Can Mama Adebayo take two Advil with her Apixaban?" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

EchoCare Kinship

by FEMZYKENTLTD

Overview Schema Source Related Servers Discussions Admin

TypeScript

Remote

🫶 EchoCare Kinship — Alexa+ MCP 2025-11-25 Guardian

Proactive Aging-in-Place, Cognitive Biomarker & Medication Guardian for Alexa+

*Mama Adebayo, 74, lives independently in Lagos. EchoCare Kinship turns her Echo Show 15 into a warm companion that quietly screens her speech for early cognitive drift, blocks dangerous medication combinations before she swallows them, calmly resolves home-safety anomalies, and keeps her family in the loop — all over a self-hosted MCP 2025-11-25 Streamable HTTP server with interactive MCP Apps (ui://), an Agent Skill (SKILL.md), and an AWS Bedrock + Strands-style multi-agent pipeline.*

Built for the "Build, Ship, Shape: Amazon Developer Hackathon" — 🎯 Alexa+ Track · 🧰 Open Source Mini-Challenge (alexa-mcp-2025-kit) · ☁️ AWS Builder Mini-Challenge

⚡ 1-Command Quickstart (zero keys, zero build step)

npm install && npm start

Then open http://localhost:3000 — the simulated Alexa+ Echo Show 15 with a live Judge & Protocol Inspector. Press ▶ Auto-Play 90s Hackathon Demo and watch the whole story, or tap any of the 4 scenario pills.

npm test        # 25 automated tests: engines, kit, live wire, 8/8 audit
npm run build   # optional: compile kit + app to dist/ (npm run start:built)

Node ≥ 20. No AWS keys needed — Bedrock runs in a seamless deterministic simulation until you set AWS_ACCESS_KEY_ID (see Going live on AWS).

Related MCP server: Sapphire Wellness MCP Server

🎬 What you'll see (the 4 profound tools)

Every tool is a real MCP tool on /mcp bound to an interactive ui:// MCP App card rendered inside the Echo Show frame. Card buttons fire genuine follow-up tools/call requests through the MCP Apps postMessage bridge.

#	MCP Tool	MCP App view	Engine	The moment that matters
🧠 1	`analyze_cognitive_checkin`	`ui://echocare/memory-lane`	Linguistic & Cognitive Biomarker Engine — MATTR lexical diversity, pronoun-to-noun drift (anomia signal), word-finding pause latency, idea density, orientation index, 7-day longitudinal z-score drift alert	A warm wedding-day reminiscence is scored against Mama Adebayo's own baseline; the "early-drift sample" shows vague "that thing" speech detected
💊 2	`verify_medication_safety`	`ui://echocare/med-guardian`	Geriatric Pharmacology & AGS Beers Criteria Engine — drug–drug interaction rules with mechanisms & modeled relative risk, anticholinergic burden (ACB)	"Can I take two Advil?" → BLOCKED: Ibuprofen+Apixaban CRITICAL ≈3.2× GI-bleed, +Lisinopril SEVERE renal, +Donepezil MODERATE — safe Acetaminophen offered & confirmable
🔥 3	`triage_ambient_anomaly`	`ui://echocare/safety-shield`	Ambient Multi-Sensor Triage — weighted risk model + 3-stage calm de-escalation state machine	Stove 375°F × 28 min, no kitchen motion → gentle check-in (60 s window) → one tap cuts the smart relay → risk re-scored to ~0, no family alarm needed
👨‍👩‍👧 4	`dispatch_care_circle_digest`	`ui://echocare/care-circle`	AWS Bedrock Runtime + Strands-style agents — CognitiveAnalyst → MedicationGuardian → SafetyTriageOfficer → DigestComposer	The whole day becomes one Peace-of-Mind digest for Tolu, Kunle & Dr. Okafor, with a live agent reasoning trace

Also exposed: 3 prompts (morning_checkin_coach, medication_safety_conversation, caregiver_handoff_brief), and the SKILL.md itself as an MCP resource (skill://echocare-kinship/SKILL.md).

🏗️ Architecture

┌───────────────────────────────────────────────────────────────────────────┐
│  Echo Show 15 Simulator (public/)  — a REAL browser MCP client + Apps host │
│  • Fetch/SSE JSON-RPC client   • MCP Apps postMessage bridge              │
│  • Judge Inspector: wire log · 1-click audit · SKILL.md/Bedrock trace     │
└───────────────▲───────────────────────────────────────────────────────────┘
                │  Streamable HTTP /mcp  (POST · GET SSE · DELETE)
                │  MCP-Protocol-Version: 2025-11-25 · MCP-Session-Id
┌───────────────┴───────────────────────────────────────────────────────────┐
│  alexa-mcp-2025-kit guard (packages/alexa-mcp-2025-kit — synced OSS copy) │
│  Origin 403 · Protocol-Version 400 · session registry · SSE event store   │
│  event ids "<sessionId>:<seq>" + Last-Event-ID replay                     │
├───────────────────────────────────────────────────────────────────────────┤
│  McpServer (@modelcontextprotocol/sdk)  +  ext-apps registerAppTool       │
│  ┌────────────────┬────────────────┬───────────────┬───────────────────┐  │
│  │ cognitive      │ medication     │ ambient       │ care-circle       │  │
│  │ biomarkers     │ guardian       │ triage        │ digest            │  │
│  │ (MATTR…)       │ (Beers, ACB)   │ (state mach.) │ (Strands agents)  │  │
│  └────────────────┴────────────────┴───────────────┴────────┬──────────┘  │
│                                        AWS Bedrock Runtime  │ Converse    │
│                                        (live w/ keys, else  │ API         │
│                                         deterministic sim)  │             │
├───────────────────────────────────────────────────────────────────────────┤
│  Agent Skill: skills/echocare-kinship/SKILL.md (loaded at boot by kit)    │
└───────────────────────────────────────────────────────────────────────────┘

Runtime technology hooks (imported and executed, per hackathon rules): @modelcontextprotocol/sdk (McpServer + StreamableHTTPServerTransport) · @modelcontextprotocol/ext-apps/server (registerAppTool/registerAppResource, RESOURCE_MIME_TYPE) · @aws-sdk/client-bedrock-runtime (BedrockRuntimeClient, ConverseCommand) · alexa-mcp-2025-kit (createStreamableHttp2025Guard, defineAlexaMcpAppTool, loadAgentSkillManifest, runMcp2025ComplianceAudit).

🧰 Open Source Mini-Challenge — `alexa-mcp-2025-kit`

Our companion open-source toolkit is github.com/FEMZYKENTLTD/alexa-mcp-2025-kit — a reusable library for building Alexa+ MCP 2025-11-25 servers: the Streamable HTTP guard (Origin/protocol/session/SSE-resumability), the MCP-Apps tool binder, the SKILL.md manifest loader, and the 8-point live compliance audit.

How this repo consumes it (synchronization rule): to guarantee npm install && npm start works 100% autonomously, the kit is vendored as a synced local workspace module at packages/alexa-mcp-2025-kit (pre-built dist/ committed) and referenced from the root package.json as:

"alexa-mcp-2025-kit": "file:./packages/alexa-mcp-2025-kit"

The ✅ 2025-11-25 Audit tab in the UI runs the kit's runMcp2025ComplianceAudit() live against this server — the toolkit proving the product, in public, on stage.

☁️ AWS Builder Mini-Challenge — Bedrock + Strands

src/agent/aws-bedrock-strands.ts implements a Strands-style agent graph (specialized agents, system prompts, shared session state, typed tools = our deterministic engines). The DigestComposer node calls Amazon Bedrock via BedrockRuntimeClient + ConverseCommand.

Going live on AWS Bedrock

export AWS_ACCESS_KEY_ID=... AWS_SECRET_ACCESS_KEY=... AWS_REGION=us-east-1
# optional: export BEDROCK_MODEL_ID=global.anthropic.claude-sonnet-4-5-20250929-v1:0
npm start   # digest now reports mode: aws-bedrock-live with token usage in the trace

With no keys (judges' default) the identical pipeline synthesizes deterministically and honestly reports aws-bedrock-simulated in /health, the UI badge, and every digest payload.

🔌 Connect your own MCP client

MCP Inspector

npx @modelcontextprotocol/inspector   # → Streamable HTTP → http://localhost:3000/mcp

Raw wire (curl) — full session lifecycle:

# 1) Initialize (note the MCP-Session-Id + SSE ids "<session>:1", ":2")
curl -i -X POST localhost:3000/mcp -H 'content-type: application/json' \
  -H 'accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-11-25","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}'

# 2) Call a tool with the session + protocol headers
curl -X POST localhost:3000/mcp -H 'content-type: application/json' \
  -H 'accept: application/json, text/event-stream' \
  -H 'MCP-Session-Id: <SESSION>' -H 'MCP-Protocol-Version: 2025-11-25' \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"verify_medication_safety","arguments":{"question":"can I take two Advil with my morning pills?"}}}'

# 3) Terminate → subsequent calls return 404
curl -X DELETE localhost:3000/mcp -H 'MCP-Session-Id: <SESSION>' -H 'MCP-Protocol-Version: 2025-11-25'

Try the guardrails: MCP-Protocol-Version: 1999-01-01 → 400; header X-Simulate-Untrusted-Origin: https://evil.example → 403; GET /mcp with Last-Event-ID: <session>:2 → missed SSE events replay.

📂 Project structure

echocare-kinship-alexa-mcp/
├── skills/echocare-kinship/SKILL.md     # Agent Skill manifest (loaded at boot)
├── packages/alexa-mcp-2025-kit/         # synced OSS toolkit (github.com/FEMZYKENTLTD/alexa-mcp-2025-kit)
│   └── src/index.ts                     # guard · ui binder · skill loader · audit
├── src/
│   ├── server/
│   │   ├── index.ts                     # Express + /mcp Streamable HTTP + REST + static
│   │   ├── state.ts                     # longitudinal day-state shared by tools
│   │   └── tools/                       # 4 domain tools + ui:// views + prompts
│   ├── engines/                         # the real algorithms
│   │   ├── cognitive-biomarkers.ts      # MATTR, pronoun drift, idea density, z-drift
│   │   ├── medication-guardian.ts       # Beers-criteria DDI rules + ACB scoring
│   │   └── ambient-triage.ts            # risk model + 3-stage de-escalation machine
│   ├── agent/aws-bedrock-strands.ts     # BedrockRuntimeClient + Strands agent graph
│   └── data/elder-profile.ts            # Mama Adebayo profile, meds, baselines, sensors
├── public/                              # Echo Show 15 simulator + Judge Inspector
├── tests/echocare-mcp.test.ts           # 25 tests (node:test)
├── DEVPOST_SUBMISSION.md                # complete copy-paste Devpost answers
└── LICENSE                              # MIT

🛡️ Safety, privacy & honesty notes

EchoCare produces wellness screening signals against a personal baseline — never diagnoses. The UI, SKILL.md guardrails, and every payload say so; blocked medications always route to a pharmacist/physician path.
Family see trends and digests, not verbatim transcripts, unless the elder taps Save to Family Album.
Demo data is deterministic and modeled (risk multipliers are labeled "modeled"). Swap in real device telemetry, pharmacy feeds, and FHIR export behind the same tool contracts.

👥 For hackathon judges (private-repo evaluation)

If this repository is private during judging, access has been granted to testing@devpost.com and to GitHub users chris-trag, knmeiss, giolaq, anishamalde, mosesroth, emersonsklar (Settings → Collaborators). To evaluate:

git clone https://github.com/FEMZYKENTLTD/echocare-kinship-alexa-mcp.git
cd echocare-kinship-alexa-mcp
npm install && npm test && npm start   # → http://localhost:3000, press ▶ Auto-Play

Submission answers, product feedback, friction log, feature requests, and the timed 2-minute demo voiceover script are in DEVPOST_SUBMISSION.md.

📜 License

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityMaintained

ResponsivenessNo issues

Resources

GitHub Repository

Need Help?

Related Servers

Related MCP Connectors

SMI Aware Assistant AOAuth
com.smiaware.mcp
Remote MCP server exposing SMI Aware tools, resources, and skills over Streamable HTTP.
Social Media Search
401
UnifAPIOAuth
com.unifapi
Hosted MCP server for live public-data APIs and Skills for AI agents.
Open Data Search AI & Machine Learning
2
Reasoning Commons
io.github.chenyuan35
AI Reasoning Cache & Consensus Layer with 11 MCP tools via Streamable HTTP.
AI & Machine Learning Knowledge & Memory Autonomous Agents
2
Word Of Clout ACN MCP Server
io.github.wordofclout
Guarded MCP server for agent-readable business truth, provenance, readiness, and discovery.
CRM Customer Data Platforms
2

Related MCP Servers

MediLinkAI
Biology & Medicine Health & Wellness
MiChaelinzo
A
license
Not graded
quality
D
maintenance
Clinical decision-support MCP server that lets AI agents reason over live FHIR patient data for medication review, appointment scheduling, and care gap identification.
Updated 4 months ago (2026-06-12 20:21 UTC)
4,113 npm
MIT
Sapphire Wellness MCP Server
Health & Wellness Fitness Tracking
J-O-WORK-REDESIGN-FDE-COHORT
F
license
Not graded
quality
C
maintenance
MCP server that exposes health metrics from the Sapphire Wellness App to AI assistants, enabling natural language queries for activity, blood pressure, glucose, heart rate, sleep, and SpO2 data.
Updated 2 months ago (2026-08-11 10:00 UTC)
-
SeniorCare MCP
Health & Wellness Calendar Management
m-khujnare2005
F
license
Not graded
quality
D
maintenance
This MCP server enables medicine management, stock/refill estimation, doctor search and availability checks, and appointment management for senior care scenarios.
Updated 2 months ago (2026-08-14 16:59 UTC)
-
Raksha
Health & Wellness Security Autonomous Agents
Aakashdeeeep
A
license
Not graded
quality
B
maintenance
A self-hosted MCP server that lets Alexa+ help care for an older family member — medicines, health readings, family messages, errands and scam calls — with every tool call screened by a deterministic safety gate and a Cedar policy. It applies zones, tripwire lockdowns, scam watches and approval-fatigue limits so money actions require explicit family approval.
Updated 18 days ago (2026-10-03 16:10 UTC)
MIT