Day.new for agents
Give AI tools a structured way to read and write the same meals, goals, workouts, sleep, habits, and daily notes you already manage in Day.new.
Day.new agent tools
private calendar actions
Find items from a menu and estimate what fits your day.
Turn a note, photo, or voice memo into an editable meal log.
Pull product nutrition and serving sizes into your log.
Update calorie, protein, and preference targets.
Use training and recovery context when choosing meals.
Pull meals, workouts, sleep, notes, and custom logs together.
Your calendar becomes a usable agent memory
The goal is not a generic chatbot. The agent should operate through typed Day.new actions and durable records that the app can show, edit, sync, and audit.
Read a day or range
Pull meals, custom log entries, workouts, sleep, readiness, and view settings into one context window.
"Show my nutrition, workouts, and sleep for last Tuesday."
Log and revise meals
Turn text, voice transcripts, or meal photos into editable itemized nutrition records with per-unit macros.
"Log a turkey sandwich and chips for lunch, then adjust the sandwich to 1.5 servings."
Track goals against reality
Compare the day against calorie, protein, and preference targets without making the agent re-learn your plan.
"How much protein do I still need today, and what dinner would keep me under my calorie max?"
Use connected sources
Let the agent reason over Oura, Hevy, Strava, and future integrations after the user connects them in Day.new.
"Was my higher hunger today more likely tied to yesterday's workout or last night's sleep?"
Manage custom tracking
Create log categories, record daily values, and keep sheet and calendar views aligned with the user's setup.
"Add a caffeine column and set today to 120 mg."
Explain the pattern
Summarize weeks of structured records into plain-language observations while preserving the source data.
"Find the meals that helped me hit protein without overshooting calories this month."
Use Day.new from your favorite agents
Connect Day.new once, then ask Claude, ChatGPT, Codex, or another MCP client to help with meals, goals, and your calendar.
Cursor
Cursor supports encoded MCP install deep links. We are not showing Cursor in this client list, but the pattern is real.
Codex
Codex supports plugin install deep links from known marketplaces. A raw MCP URL still belongs in the CLI or config.toml unless Day.new ships a Codex plugin.
Claude Desktop
Claude Desktop one-click install is through a packaged desktop extension. Until Day.new ships one, JSON config is the right install path.
ChatGPT
ChatGPT custom MCP connectors are created in connector settings or developer mode; there is no documented public add-connector deep link.