Use these docs with your AI assistant - Groovin

An AI assistant writes much better code when it reads the documentation first. These docs are built for that.

Give it the docs

The whole site: point your assistant to https://docs.groovin.ai/llms.txt, which lists every page.
One page: add .md to any page address to get its plain text. For example: https://docs.groovin.ai/reference/limits.md.

Paste these rules first

Your assistant does not know what is risky unless you tell it. Paste this at the start of your conversation:

You are helping me use the Groovin API. Follow these rules:

- Every call acts on my real LinkedIn account and my real CRM. There is no test mode.
- Before writing code, read https://docs.groovin.ai/reference/side-effects.md
  and https://docs.groovin.ai/reference/limits.md
- Never write a loop that sends LinkedIn messages or invitations.
- Never loop get_linkedin_profile over a list of people.
- Leave random gaps between LinkedIn calls. Never use a fixed schedule.
- Never retry a send automatically. After an error or a timeout, check first.
- On a 429, stop and wait at least 5 minutes.
- Read my API key from an environment variable. Never print it or log it.
- Stop and ask me before running anything that writes.

Check what it wrote

Before running a script, look for these things, or ask your assistant to point them out:

Look for	Why it matters
A loop around `linkedin/messages` or `linkedin/invitations`	Bulk sending can get your LinkedIn account restricted
A loop around `linkedin/profile`	Each call is a real profile view, limited to about 100 a day
An automatic retry on a send	The same person can get the message twice
`linkedin/relations` without `limit`	It reads your whole network from LinkedIn
`enabled: false` on `crm/contacts/sync-toggle`	It deletes conversation notes in your CRM
Your API key written in the code	Anyone who sees the file can use your account

Using the MCP instead

With the MCP, your assistant calls Groovin itself. Each tool tells it what the tool does and when to ask you first. Whether it asks you before an action that writes depends on your AI tool: check its settings before you start.