A2Me MCP Server
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@A2Me MCP ServerWhen is grandma's birthday?"
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.
A2Me MCP Server
v1 · read-only. A Model Context Protocol server that exposes read-only, privacy-redacted family-context tools for the A2Me family social platform — so assistants like Claude, ChatGPT, or KAI can answer "when is mom's birthday?" or help write a message to grandma, scoped to your own family.
Hosted at https://mcp.a2me.app/mcp and used in production for A2Me's Connect a2me feature.
Connect from an AI assistant
In Claude or ChatGPT, add a custom connector using the URL:
https://mcp.a2me.app/mcpYou'll be sent through OAuth to sign in to A2Me and grant read-only access; the assistant then has family-aware context scoped to your account.
Related MCP server: telegram-mcp-server
Run it locally
npm install
npm run dev # stdio transport, mock data by defaultIt runs on stdio by default (for local MCP clients) and supports a remote HTTP transport (MCP_TRANSPORT=http) for the hosted deployment. By default it uses mock data (A2ME_USE_MOCK=true); set A2ME_API_URL + A2ME_USE_MOCK=false to call a real API. See .env.example.
Available Tools
Tool | Description |
| Returns the user's family members with relationship labels |
| Birthdays, anniversaries, and events in the next N days |
| Recent posts, photos, videos, birthday cards |
| A family member's profile, interests, and activity summary |
| How two family members are related |
| Context for writing a birthday card (memories, interests, tone) |
| Fuzzy search by name or relationship ("mom", "my grandmother") |
| Natural language date questions ("When is mom's birthday?") |
| Context for writing a message with suggestions |
Prompts
The server also exposes prompts — one-click starting points that appear as connector commands in Claude (and suggestions in ChatGPT), so users get value without knowing tool names. Each steers the assistant to the read-only tools above:
Prompt | What it does |
| Draft a birthday card grounded in a person's context |
| Draft a message for any occasion/tone |
| Summarize what's new with the family |
| List upcoming birthdays/anniversaries with suggestions |
| Warm summary of a family member and how you're related |
The person argument on these prompts autocompletes from your family roster
(names + relationship labels) as you type — MCP argument completion, scoped to your
family. See src/prompts/index.ts and
src/completions.ts.
Example Scenarios
"Help me write a birthday card for my sister"
find_family_member→ resolves "my sister" to Sarah Walkerget_birthday_card_context→ returns interests (painting, hiking, coffee), tone suggestions
"When is grandma's birthday?"
answer_family_date_question→ returns Margaret Walker's birthday (November 8)
"What's been happening in the family?"
get_recent_family_activity→ returns recent posts, photos, events
"Help me write a thank you message to my dad"
get_message_context_for_person→ returns Robert Walker's context, interests, suggestions
"Who's in my family?"
get_family_members→ returns all family members with relationships
"What events are coming up?"
get_upcoming_family_dates→ returns upcoming birthdays, anniversaries, events
Architecture
┌─────────────────────────────────────────────────────────┐
│ MCP Client (LLM) │
│ (Claude, ChatGPT, KAI, etc.) │
└───────────────────────┬─────────────────────────────────┘
│ stdio (MCP Protocol)
┌───────────────────────▼─────────────────────────────────┐
│ A2Me MCP Server │
│ ┌─────────────────────────────────────────────────┐ │
│ │ Tool Registry (9 tools) │ │
│ └──────────────────────┬──────────────────────────┘ │
│ │ │
│ ┌──────────────────────▼──────────────────────────┐ │
│ │ Family Context Resolver │ │
│ │ (fuzzy matching, relationship resolution) │ │
│ └──────────────────────┬──────────────────────────┘ │
│ │ │
│ ┌──────────────────────▼──────────────────────────┐ │
│ │ A2Me API Client │ │
│ │ (currently mock, future: real HTTP calls) │ │
│ └──────────────────────┬──────────────────────────┘ │
│ │ │
│ ┌──────────────────────▼──────────────────────────┐ │
│ │ Auth Context │ │
│ │ (currently mock, future: OAuth/Firebase) │ │
│ └─────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────┘
│ (future)
┌───────────────────────▼─────────────────────────────────┐
│ A2Me REST API │
│ (kinnectd-api service) │
└─────────────────────────────────────────────────────────┘Authentication
Currently uses mock authentication — always returns an authenticated user ("Alex Walker") for local development.
Future production integration:
OAuth2/Firebase token validation via A2Me API
Token passed via MCP session context or environment
Per-user data isolation (users only see their own family)
Required A2Me API Endpoints (Future)
Method | Endpoint | Description |
GET |
| List user's family members |
GET |
| Upcoming dates with filters |
GET |
| Recent family activity feed |
GET |
| Person profile with safe context |
GET |
| Relationship path between two people |
GET |
| Birthday card writing context |
ChatGPT app widgets
For ChatGPT (Apps SDK), four tools render an interactive inline widget instead of plain JSON:
get_upcoming_family_dates→ an Upcoming family dates cardget_family_members→ a Your family rosterget_person_profile→ a Family member profile cardget_recent_family_activity→ a Recent family activity feed
Each widget is a small React bundle in widgets/src/ built by
npm run build:widgets into dist-widgets/<name>.js|.css. The server serves those as
static assets (/widgets/...) and exposes each as a ui://widget/<name>.html
resource (MIME text/html+skybridge); the paired tool carries
_meta["openai/outputTemplate"] pointing at it. The widget reads the tool's
structuredContent from window.openai.toolOutput and renders it. Other MCP clients
(Claude, KAI) ignore the widgets and use the same tools' text output. See
src/widgets/registry.ts.
Development
npm run dev # Run the server with tsx (hot reload)
npm run build # Compile server (tsc) + build widget bundles
npm run build:server # Server only
npm run build:widgets # Widget bundles only (-> dist-widgets/)
npm run test # Run tests (incl. widget render tests)
npm run test:watch # Watch mode
npm run lint # ESLint
npm run format # Prettier
npm run check # Type check onlyPrivacy Design
This server is designed to be privacy-first:
No email addresses, phone numbers, or physical addresses are ever returned
Birthdays are shown as month-day only (no birth year)
No financial or health information
All data scoped to the authenticated user's family only
Managed accounts (children) have additional protections
Privacy Policy
Privacy policy: https://a2me.app/privacy
This connector accesses A2Me data on behalf of the authenticated user, over an OAuth 2.0 "Connect A2Me" flow, and is bound by the A2Me privacy policy above.
What we collect / access: read-only family-context data for the authenticated user's own family — member names, relationship labels, month–day of birthdays and events, and recent activity summaries. We never return email addresses, phone numbers, physical addresses, birth years, or financial/health data.
How it's used: returned to the connected AI assistant solely to answer the user's request in-session. The connector does not train models on this data and performs no writes back to A2Me.
Storage & retention: the connector holds no family data at rest. For transparency and abuse prevention we log access metadata (timestamp, tool name, calling assistant, scopes) in an append-only audit log; users can review this under Settings → Connected apps in A2Me.
Third-party sharing: OAuth tokens are issued and validated via our auth provider (Scalekit); no family data is shared with third parties beyond the AI assistant the user explicitly connected.
Contact: privacy@a2me.app · security disclosures per SECURITY.md.
Tech Stack
Runtime: Node.js 20+
Language: TypeScript (strict mode)
MCP SDK:
@modelcontextprotocol/sdkValidation: Zod
Testing: Vitest
Transport: Streamable HTTP (remote/production) and stdio (local dev)
License
Private — Kinnectd / A2Me
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Alicense-qualityDmaintenanceA read-only MCP server that enables AI assistants to access Slack workspace data, including channels, messages, threads, and user information.Last updated43,1793MIT
- Alicense-qualityCmaintenanceA read-only MCP server that lets AI agents read personal Telegram chats from an allowlist of folders, with no send/edit/delete capability.Last updated28MIT
- Alicense-qualityDmaintenanceMCP server for ContextKeeper that enables AI assistants to read and modify context items through natural language commands.Last updated6MIT
- Alicense-qualityBmaintenanceEnables AI assistants to query local conversation transcripts and relationship data via a local MCP server, allowing retrieval of past conversations and facts about people.Last updated1AGPL 3.0
Related MCP Connectors
The personal context layer for AI: your profile and files, read by any MCP client over OAuth.
Pocket Agent (aipocketagent.com) MCP server — read tools for personas, apps, and product info.
Person-owned, portable AI memory as a remote MCP server, readable and writable by any MCP client.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Kinnectd/a2me-mcp-server'
If you have feedback or need assistance with the MCP directory API, please join our Discord server