mcpField
D
Dayze

Dayze

Remote
dayze.comMemory & Knowledge0Updated 1w ago

About Dayze

Dayze provides AI agents with personal life context through contacts, plans, and memories. It separates private records from a public notable-people catalog, and supports source tracing and corrections.

Key features

  1. 01Read personal life context
  2. 02Find contacts and related events
  3. 03Inspect sources and correct records
  4. 04Search the public notable-people catalog
  5. 05Authenticate with OAuth

Use cases

  1. 01Ask about recent plans and memories
  2. 02Find interactions linked to a contact
  3. 03Verify and correct the basis of an answer

MCP support

Dayze offers a hosted MCP server at https://dayze.com/api/mcp for compatible clients to read and manage life context. The compact profile is read-only; public notable-people tools do not require sign-in.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://dayze.com/api/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://dayze.com/api/mcp
Streamable HTTPhttps://dayze.com/api/mcp/key

Tools

15
ParameterTypeRequiredDescription
qstringYesSearch query (alias: query)
limitnumberNoMax results 1–20

Details

From the project README

Dayze MCP

Dayze MCP

Dayze gives connected AI tools Life Context: your people, plans and memories in a private, correctable record.

After you sign in, get_context_pack hands your AI the signed-in user's context (who matters, this week's plans, recent memories) over the Model Context Protocol (MCP). The public notable-people catalog (notable_*) is a separate feature and never touches your private record.

ChatGPT + iPhone first. Capture people, plans and moments in Dayze on your iPhone, then connect ChatGPT and ask about your own life. Cursor, Codex, Claude and other MCP clients use the same hosted endpoint.

Listed on LightNow: https://lightnow.ai/servers/com.dayze/life-context

LightNow MCP capabilities

Relationship intelligence

Dayze starts with the people in your life. Ask "When did I last see Maya, and what's next with her?" and your AI can:

  1. Resolve the person — resolve_person / get_people against your private Dayze Contacts.
  2. Show the linked moments — get_person_interactions returns the events, places and notes connected to that person.
  3. Show the evidence — explain_fact returns the source behind a claim, so the answer is traceable rather than guessed.
  4. Correct it — update_person, merge_people and the other write tools fix the record, so the next answer is right.

These four steps use the full endpoint (https://dayze.com/api/mcp). ChatGPT's compact profile is read-only: get_context_pack, get_people_context, get_events, search_dayze and related context reads, with corrections made in Dayze itself.

Connect ChatGPT

  1. ChatGPT → Settings → Connectors → add a custom connector.
  2. URL: https://dayze.com/api/mcp?tools_profile=compact
  3. Click Connect and sign in to Dayze (OAuth). No API key.
  4. Ask: "Use Dayze to pull my context pack for this week."

More detail: README-CHATGPT.md · https://dayze.com/docs/agents

Claude Code

This repo is also a Claude Code plugin and its own marketplace (.claude-plugin/). The plugin adds the hosted Dayze MCP server and the dayze-life-context skill.

claude plugin marketplace add gohluke/dayze-mcp
claude plugin install dayze@dayze
claude mcp login plugin:dayze:dayze

The last command opens Dayze sign-in in your browser (OAuth). In a session you can use /mcp → plugin:dayze:dayze → Authenticate instead. No API key.

Then ask: "Use Dayze to pull my context pack for this week."

To point the plugin at a different Dayze connector URL, set DAYZE_MCP_URL in the env block of ~/.claude/settings.json and sign in again. The default is https://dayze.com/api/mcp.

Local development: claude --plugin-dir /path/to/dayze-mcp.

Cursor / Grok Bot

This repo is a Cursor Plugin (v1.33.0): .cursor-plugin/plugin.json, skills/dayze-life-context/, URL-only root mcp.json, plus Codex/ChatGPT metadata in .codex-plugin/ and .mcp.json.

Dayze Contacts is the private CRM. Advertised MCP names stay get_people, create_person, update_person, resolve_person (stable). Call aliases get_contacts / create_contact / update_contact / resolve_contact work on the hosted server. Public People stays notable_*.

Cursor Marketplace: https://github.com/gohluke/dayze-mcp submitted 2026-09-19, in review (see PUBLISH.md). The ChatGPT custom connector works today; the ChatGPT Apps Directory listing is in review.

Until marketplace review lands, test locally:

mkdir -p ~/.cursor/plugins/local
ln -s /path/to/dayze-mcp ~/.cursor/plugins/local/dayze

Reload Cursor (Developer: Reload Window), open Customize, and click Connect. Sign in to Dayze in the browser. Do not paste an API key — Cursor discovers OAuth from RFC 9728. Same Connect flow for Cursor desktop, Cursor web, and Grok Bot.

Docs: https://dayze.com/docs/agents

Websitehttps://dayze.com
Agents docshttps://dayze.com/docs/agents
Streamable HTTPhttps://dayze.com/api/mcp
REST MCPhttps://dayze.com/api/mcp
Discoveryhttps://dayze.com/.well-known/mcp.json
Server cardhttps://dayze.com/.well-known/mcp/server-card.json
OpenAPIhttps://dayze.com/openapi.json
OAuth PRMhttps://dayze.com/.well-known/oauth-protected-resource
OAuth AShttps://dayze.com/.well-known/oauth-authorization-server

Quick try

# Streamable HTTP (JSON-RPC)
curl -X POST https://dayze.com/api/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"curl","version":"1.0"}}}'

curl -X POST https://dayze.com/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'

# REST (compat)
curl https://dayze.com/api/mcp

curl -X POST https://dayze.com/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"tool":"notable_pack","parameters":{"slug":"albert-einstein"}}'

The last call uses the public notable-people catalog. Timeline events include day_number (e.g. Einstein’s Nobel = Day 15,580).

Auth

  • Private Life Context tools (get_context_pack, Contacts, events, memories): OAuth dayze_at_… (recommended) or Bearer dayze_k_…, scope context
  • Public notable_* tools: no login (see public tool pricing)
  • OAuth 2.1 + PKCE + DCR for ChatGPT / Claude / Gemini agents — see https://dayze.com/docs/agents

Transport

  • Streamable HTTP JSON-RPC at /api/mcp (initialize, tools/list, tools/call)
  • Compact ChatGPT profile at /api/mcp?tools_profile=compact (read-only subset)
  • REST MCP-compatible at /api/v1/mcp (GET capabilities, POST {tool, parameters})
  • GET /api/mcp returns discovery JSON (200); SSE sessions are not available on Netlify serverless

Tags

life-context · mcp · model-context-protocol · personal-crm · relationship-intelligence · chatgpt · ai-agents · notable-people

Glama install / Make Release

Dayze MCP is hosted at https://dayze.com/api/mcp. This repo ships a local stdio adapter (server.mjs) so Glama can build/scan without putting a URL in CMD (Glama rejects remote endpoints in CMD arguments).

  1. Open https://glama.ai/mcp/servers/gohluke/dayze-mcp/admin/dockerfile
  2. Build steps: ["npm install"]
  3. CMD arguments: ["node", "./server.mjs"]
  4. Click Build → wait for green → Build & Release (1.6.2)

If Glama keeps checking out an old commit, use build steps: ["git fetch origin && git checkout origin/main", "npm install"]

Prefer connecting clients directly to https://dayze.com/api/mcp (Streamable HTTP + OAuth).

Technical reference: public tool pricing

This applies only to the public notable-people catalog, not to your private Life Context.

License

Documentation and listing metadata in this repo: MIT. The Dayze product and API remain proprietary; this repo exists so directories can index a public GitHub URL.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.