Teach your agent the HeatCheck playbook.
An installable skill teaches your AI assistant how to work the heat: who to surface, how to explain a score, and when to draft a task.
The problem
A raw API is not a workflow. An agent with tools but no playbook produces noise, not the crisp shortlist a rep wants.
What it does
Behavioral rules
Golden rules first, including “every heat score is a receipt, not a vibe.”
Playbooks
When to surface a crossing, how to explain it, how to draft.
Hand-back pattern
The skill ends every task by handing a link back into HubSpot.
How it works
Install
Drop the skill into your agent’s skills directory.
Connect
Pair it with the MCP server.
Work
The agent follows the playbook against your heat data.
The skill routes value through the scoring tools and always ends in a human handoff, never an autonomous send.
A tool without a playbook is noise. The skill is the playbook.
The short version.
What format is the skill?
A Markdown skill you install in your assistant’s skills directory.
Does it need the MCP server?
Yes, the skill drives the MCP tools; together they turn the agent into a heat analyst.
Can it act on its own?
It drafts and hands back. Sending stays a human decision by design.
See who’s hot right now.
Early access is open for a small design-partner group.