Streamliner One
Polycal by Streamliner One · Early Access

Polycal — Calendar aggregation for agents

One merged, read-only calendar across every Google and CalDAV account you connect — exposed as an MCP server your agent can query, plus a safe local layer where your agent can remember what each calendar means.

Available as part of Streamliner One managed AI assistant installations.

Humans have Fantastical.
Agents have nothing.

People with calendars scattered across many accounts use desktop apps like Fantastical, BusyCal, or Apple Calendar to merge them into one coherent view. Agents do not have an equivalent.

Existing MCP servers are single-account

There are MCP servers for Google Calendar and for CalDAV — but each one talks to one account at a time. They do not compose into the multi-account picture that real life looks like.

Real life is multi-account, multi-provider

Many Google Workspace accounts with shared calendars. A few Fastmail accounts other people are also writing to. A family iCloud calendar. No single MCP server sees the union.

Polycal is the missing layer: one read-only calendar projection across every account you connect, available to any MCP-compatible agent.

What Polycal Does

Polycal is built for people who use multiple calendars across Google, Fastmail, iCloud, and other CalDAV providers, and want their agents to understand availability without relying on a desktop calendar app being open.

It runs alongside a managed Streamliner One AI assistant installation. Users connect calendar accounts once, select which calendars are visible, and Polycal keeps a local read-only calendar projection that agents can search and query through an API and an MCP endpoint.

And it does more than merge events. Polycal gives the agent a local layer to remember what each calendar is for — so it doesn't just see what's on your calendar, it understands the system underneath.

Multi-account connectors are one of Claude's most-requested missing features. Most agents, MCP servers, and chat clients still talk to one calendar account at a time. Polycal is what fills that gap.

Main Features

Every account, merged

Many Google Workspace accounts, multiple Fastmail, a family iCloud — all merged into one calendar layer your agent can reason over. Google via managed OAuth; CalDAV via app-specific passwords.

Per-calendar visibility

Choose exactly which calendars Polycal sees. Everything else stays hidden from Polycal and from your agent.

Agent-maintained calendar context

Aliases, tags, notes, semantic types, agent-facing names. Maintained by your agent, correctable by you.

Honest free/busy

Events marked transparent stay transparent. A flight on your calendar as a reference does not look like a meeting to your agent.

Merged search and free-slot detection

Search and find free slots across every calendar from every account you connected — locally, in real time. No staring at a UI.

MCP + local API

Any MCP-compatible client plugs in instantly. No glue code. No per-provider servers to maintain.

What you can finally ask your agent

Things that were impossible across many accounts before, asked in plain language.

"What's my actual availability across everything next week?"

"Find a 30-minute slot when I'm free across all my work calendars."

"Don't book over my partner's events on the family iCloud."

"What does tomorrow look like — combined view, please."

"Block my evenings when the kids have activities."

"When am I free, ignoring transparent events?"

Your agent answers using Polycal's merged view of every calendar you connected — locally, in real time.

Calendar context, for agents

Your agent will love it

A merged calendar is useful. A merged calendar your agent can read like a map of your life is something else.

Provider calendar names are written for humans, legacy workflows, and individual accounts. Agents inherit those names without context — so the meaning of each calendar usually lives partly in your head, partly in a chat, partly in the agent's working memory. None of those are durable.

Polycal gives the agent its own safe local layer to remember what each calendar means — aliases, tags, notes, semantic types, agent-facing names. The agent maintains this on its own, as a side-effect of working with you. Provider calendars and provider events are never modified.

Search benefits immediately. Polycal returns event matches and calendar / account / alias / tag matches separately, so instead of a misleading "nothing found," the agent can say: "No event title matched, but this project calendar did." A small distinction that quietly removes a whole class of wrong answers.

What that looks like

provider_name: Project X — Infra
agent_display_name: Project X Infrastructure
aliases: Project X Infra, Project X full name
tags: project-x, infrastructure, ops
semantic_type: project_calendar

The agent maintains this. You can correct it in the dashboard if it ever drifts. The provider calendar name on the other end stays untouched.

The dashboard you see.
The MCP your agent uses.

Polycal is not just a headless service. The same merged calendar your agent queries is rendered in a small read-only dashboard inside your managed installation, so you can verify exactly what your agent is seeing — across every account you connected.

The dashboard also exposes the semantic context your agent maintains about each calendar. Inspect it, correct anything that drifted, move on. Oversight, not data entry.

Same data. Two surfaces.

One coherent calendar — for your eyes and for your agent.

Dashboard (for you)

A read-only merged calendar view. See every selected calendar from every connected account in one place. Toggle which calendars are visible to Polycal and to agents.

MCP + API (for your agent)

A local MCP endpoint and a local API any MCP-compatible client can call. Search, merged views, and free-slot detection — without leaving your managed installation.

How It Works

1

Connect accounts

Connect Google Calendar accounts through Streamliner One managed OAuth, and CalDAV accounts (iCloud, Fastmail, others) using app-specific passwords.

2

Choose visible calendars

Pick exactly which calendars Polycal should sync. Anything you do not select stays hidden — from Polycal and from agents.

3

Polycal builds the local projection

Polycal keeps a read-only local projection of the calendars you selected, with free/busy status preserved and merged across accounts.

4

Agents query through MCP or API

Any MCP-compatible client queries calendar context locally — search, merged views, free-slot detection — without your data leaving the installation.

5

Agents remember what calendars mean

As the agent works with your calendars, it records what each one is for — aliases, projects, semantic types — locally inside Polycal. Future questions get smarter answers. Provider calendars and events are never touched.

6

Humans review when needed

The dashboard shows the same metadata your agent maintains. Correct anything that drifted; the rest takes care of itself. Oversight, not data entry.

Works with any MCP-compatible LLM harness

Polycal exposes a local MCP endpoint. If your harness speaks MCP, it can query your merged calendar.

OpenClaw
Hermes
Claude Code
Claude Desktop
Codex
ChatGPT

…and any other MCP-compatible client.

Connect accounts in seconds, not days

Getting an agent to read your Google calendars today means walking the full Google developer workflow — once per account. With Polycal, we did that work, so you don't.

You skip

  • Creating Google Cloud projects
  • Setting up OAuth apps and consent screens per account
  • Generating, downloading, and managing JSON credentials
  • Reconciling events across accounts in your agent's memory

Your data, your machine

Polycal runs where you run it. It pulls calendars from Google, iCloud, Fastmail, and other CalDAV providers into one merged read-only view — the same coherent picture Fantastical gives your eyes. The difference is what reads it: your agent, through MCP, instead of you, through a desktop app.

Streamliner One never sees your calendar data. We don't know which accounts you connected, which calendars are visible, or what's on them. The merged calendar lives on your machine, and your agent queries it locally.

And it's read-only with respect to your provider calendars on purpose. Polycal can read the calendars you choose. It does not create, edit, delete, share, or transfer events in Google Calendar, iCloud, Fastmail, or any other provider.

Your agent's local notes on what each of those calendars means stay inside Polycal too — in your managed installation. That's how your agent learns your calendar system over time, and none of it travels back to providers.

Why we built it

I run my life across roughly ten Google Workspace accounts, three Fastmail accounts, and a family iCloud calendar. Every desktop calendar app handles this for humans. Nothing handles it for agents — and the closest workaround is to spin up your own Google Cloud project, your own OAuth app, and your own consent screen, just to let your assistant see one account.

That's not a workflow. That's a punishment. So I built the thing I wanted: one MCP server, every account, set up once.

Then, after using it for a while, the next thing I wanted was for the agent to remember what each of those calendars actually meant — not relearn it conversation after conversation. So we added a place for that, too. Polycal now gives the agent something to remember with.

Availability

Polycal is in early access

It is currently offered as part of Streamliner One managed AI assistant installations, scoped per setup rather than priced as a standalone product.

Broader distribution is something we are considering — not something we are committing to today. If you would like to use Polycal inside a managed installation, get in touch and we will scope the fit.

Frequently Asked Questions

Why not just use Claude Desktop's Google integration?
Claude Desktop's native Google Calendar integration connects one account at a time. If your life lives across multiple Workspace accounts, Fastmail accounts, or a family iCloud calendar, you can't see them all coherently — and multi-account support is one of the most-requested features in the Claude community. Polycal is the multi-account upgrade: connect every account once, and Claude Desktop (or any MCP client) queries the merged read-only calendar through Polycal's local MCP endpoint.
Why not just use Fantastical?
Fantastical is for your eyes; Polycal is for your agent. Both merge calendars from many accounts into one coherent view, but Fantastical renders that view in a desktop UI a human reads, while Polycal exposes it through MCP so any agent can search, find free slots, and reason about availability — without you having to keep a calendar app open.
Which agents and LLM harnesses can use Polycal?
Anything that speaks MCP. Polycal exposes a local MCP endpoint and a local API, so it works with OpenClaw, Hermes, Claude Code, Claude Desktop, Codex, ChatGPT, and any other MCP-compatible client.
Which calendar providers does Polycal support?
Google Calendar via OAuth, and CalDAV providers such as iCloud and Fastmail using app-specific passwords. Microsoft Calendar is not supported in the current release.
Can I control which calendars Polycal sees?
Yes. After connecting an account, you pick exactly which calendars Polycal should sync. Calendars you don't select stay hidden from Polycal and from any agent querying it.
Does Polycal create, edit, or delete calendar events?
No. Polycal is read-only with respect to provider calendars. It does not create, edit, delete, share, or transfer events in Google Calendar, iCloud, Fastmail, or any other provider. Your agent can maintain local semantic notes about your calendars inside Polycal — see the next FAQ — but those stay in your managed installation and never modify provider calendars or events. Write access to providers may come later as a separate, opt-in permission.
What is agent-maintained calendar context?
It's a local semantic layer inside Polycal where your agent records what each calendar means: aliases, tags, notes, semantic types, and agent-facing names. The agent maintains it as a side-effect of working with your calendars; you can inspect and correct it in the dashboard if anything drifts. None of it modifies provider calendars or events — it's just how the agent remembers your calendar system between conversations, and how Polycal can answer "no event matched, but this project calendar / alias / tag did" instead of falsely returning nothing.
What Google Calendar permissions does Polycal request?
Two read-only scopes: https://www.googleapis.com/auth/calendar.calendarlist.readonly and https://www.googleapis.com/auth/calendar.events.readonly. They let Polycal list the calendars you can access and read event metadata from the calendars you explicitly choose to sync. Nothing else.
Is Polycal available outside Streamliner One installations?
Not yet. Polycal is currently part of Streamliner One managed AI assistant installations during early access. Broader distribution is something we're considering, not something we're committing to today.

Give your agent a calendar that actually knows your week

Polycal is currently available with Streamliner One managed AI assistant installations. Get in touch to scope a fit.

Contact Streamliner One