Dayze MCP: Connect Your Life Context to ChatGPT, Gemini, and Cursor

Chatbots are great at language. They are terrible at your life — unless you paste the same birthday, last conversation, and meal log into every new thread.

Dayze MCP is how we fix that. Dayze exposes a Model Context Protocol server so agents in ChatGPT, Google Gemini, and Cursor can call structured tools for life-in-days knowledge and — with your permission — your private Dayze graph.

This post is the practical overview: what MCP is here for, what works without login, how each client connects, and where to go next.

Why life context belongs outside the chat window

Most “AI memory” products try to remember you inside one vendor’s chat history. That breaks the moment you switch apps, start a new chat, or ask a coding agent the same question you asked ChatGPT yesterday.

Dayze takes the opposite approach:

1. You own a durable life graph — people, events, food, places, inventory, photos — on Dayze.
2. Agents fetch context when they need it over MCP (get_context_pack, resolve_person, notable_pack, …).
3. Public knowledge stays free to try — notable-people packs with day-of-life numbers, with a free daily allowance for agent traffic.

That split matches how people actually work: celebrity “how old is X?” curiosity is acquisition; the product is your timeline talking to the agents you already use.

What Dayze MCP is

Public tools (no Dayze login): notable_search, notable_profile, notable_pack, notable_pack_premium, compare_lives — structured packs for famous lives indexed by days lived.

Private tools (OAuth or API key): get_context_pack, people/CRM reads and writes, food, events, expenses, inventory, photos, and more. Unauthenticated private calls return 401 with OAuth discovery metadata so ChatGPT / Claude / Gemini can complete the browser Allow flow.

Scopes you will see: openid, email, mcp, context, offline_access.

ChatGPT — life context as a connector

OpenAI’s Apps / MCP connector path expects OAuth 2.1 + Dynamic Client Registration, not a pasted API key in the UI.

What to use

• MCP URL (recommended for ChatGPT’s tool-schema size limits):
  https://dayze.com/api/mcp?tools_profile=compact
• Full catalog for portal scans / power users: https://dayze.com/api/mcp

Try prompts like

1. Use Dayze to pull my context pack for this week.
2. Fetch the Dayze notable pack for albert-einstein and tell me what day of his life the Nobel Prize was.
3. Who is [someone] in my Dayze contacts? Summarize recent interactions.

Sign in with Dayze when ChatGPT prompts OAuth. Private writes (log a meal, update a person) stay in your Dayze workspace — not the public internet.

Developer packaging and submission notes live under our agents docs; the public pitch page is dayze.com/mcp.

Google Gemini — same MCP URL, Google OAuth handshake

Gemini Custom Apps / MCP clients use the same Streamable HTTP endpoint. Dayze’s authorization server allowlists Google’s MCP redirect prefixes and origins so the browser callback can complete.

Pattern

1. Point the Gemini MCP client at https://dayze.com/api/mcp.
2. Complete OAuth Allow on dayze.com.
3. Call public notable_* tools anonymously, or private tools after context scope is granted.

If a custom app stalls after the callback, re-check Advanced Settings (client ID/secret from Dynamic Client Registration) and that the redirect URI is the Google-issued MCP callback — not a hand-edited localhost URL.

Cursor — coding agents with your life graph

Cursor already speaks MCP. Dayze is usable today without waiting on a marketplace listing:

fetch https://dayze.com/ai to connect Life Context

Or add a remote MCP server pointing at https://dayze.com/api/mcp (OAuth; no API key committed in the repo). The public stub / plugin packaging lives at github.com/gohluke/dayze-mcp.

Why Cursor matters for Dayze

• Agents writing code about your product can pull real inventory, people, or trip context instead of inventing fixtures.
• The same get_context_pack bootstrap works in IDE agents and chat agents — one graph, many clients.

Public packs vs private graph (don’t confuse the two)

We are honest about thin celebrity pages elsewhere on Dayze: they are acquisition, not proof. MCP makes that split machine-readable — agents can demo with notable_pack, then deepen into your timeline once you sign in.

Quick start checklist

1. Open dayze.com/mcp or dayze.com/ai.
2. In ChatGPT / Gemini / Cursor, add MCP server https://dayze.com/api/mcp (ChatGPT: prefer ?tools_profile=compact).
3. For private tools, complete OAuth Allow.
4. Ask for a context pack or a notable pack — those two calls teach most agents the surface area.

# Health / version
curl -sS https://dayze.com/api/mcp/health

# Public tools list (JSON-RPC)
curl -sS -X POST https://dayze.com/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

What this is not

• Not a generic “proxy every API on the internet” MCP mesh.
• Not chat-history memory locked to one vendor.
• Not a replacement for building your private timeline in the Dayze app — MCP reads and writes that timeline from agents.

If you want the human product first: sign up free, add your birthday, and log a few people. Then connect ChatGPT, Gemini, or Cursor and ask the agent what it can see.

Frequently Asked Questions

What is Dayze MCP?

Does ChatGPT MCP need an API key?

Can I use Dayze MCP with Cursor today?

What is the difference between notable_pack and get_context_pack?

Is there a free tier for agents?