Automation, Sync, And AI Agents FAQ

These are the power-user questions. Use this page when you are deciding whether PrimeCal should do the work automatically, sync it from elsewhere, or let an AI agent act on your behalf.

Should I solve this with Automation or with an AI agent?

Short answer: use Automation for repeatable in-product rules; use an AI agent when an external tool needs controlled access to PrimeCal.

Choose Automation when:

• the trigger is predictable
• the rule should run the same way every time
• the logic lives naturally inside PrimeCal

Choose AI Agents (MCP) when:

• an external coding tool or assistant needs access
• permissions must be scoped tightly by feature or calendar
• a human or tool outside PrimeCal is initiating the work

Can imported events trigger automations?

Short answer: yes, imported events can participate in automation when you set the rule up for that workflow.

That makes a strong combination for cases like:

• recoloring imported school calendars
• creating follow-up tasks from imported events
• normalizing titles or descriptions after sync

Start small and verify one real example before building a larger rule set.

I connected Google or Microsoft. What should I sync first?

Short answer: start with one or two calendars that you genuinely need, not your whole account.

The safest first connection is a small, meaningful set such as:

• one shared family calendar
• one school or work calendar

This makes it easier to catch naming, color, duplication, and recurrence issues before the setup gets wide.

A synced calendar looks duplicated or messy. What is the safest fix?

Short answer: simplify first, then reconnect cleanly if needed.

Work in this order:

1. confirm which calendars are actually mapped
2. reduce the connection to the smallest useful set
3. re-check whether two-way behavior is appropriate
4. if the mapping is wrong, disconnect and reconnect cleanly instead of stacking more changes on top

Can an AI agent read my whole account by default?

Short answer: no. PrimeCal agents are meant to be permissioned and scoped.

The safest approach is to grant:

• only the actions the tool needs
• only the calendars or automation rules it needs
• only one key per tool or workflow

What is the safest first test after creating an agent?

Short answer: test one low-risk read or one low-risk write against a non-critical calendar.

Good examples:

• list events from one test calendar
• create one test task
• trigger one non-destructive automation rule

Do not start with a broad write scope or a production-critical calendar.

Do I need one agent per tool?

Short answer: yes, in most cases that is the cleaner and safer pattern.

Separate agents make it easier to:

• understand who or what the key belongs to
• revoke one client without affecting others
• narrow permissions accurately

Can I combine sync, automation, and AI agents?

Short answer: yes, but layer them in that order of stability.

Best-practice rollout:

1. get the external sync result correct
2. add one automation rule
3. add an AI agent only after you understand the stable data shape

Where should I go next?

• Introduction To Automation
• Managing And Running Automations
• External Sync
• Agent Configuration