Create Your First Playbook

By the end of this guide, you'll have a published playbook your AI tools can use.

1. Create a new playbook

Go to your workspace and click New Playbook. You'll be asked to choose a type — select Prompt for most workflows. A prompt playbook is a reusable instruction set your agents follow whenever they run it.

Fill in three fields to get started:

• Title — a short, descriptive name (e.g. "Meeting Summary")
• Summary — one sentence explaining what it produces (e.g. "Summarises meeting notes into action items and key decisions")
• Function area — the team or discipline this belongs to (e.g. Marketing, Operations, Customer Success)

Hit Create and you'll land in the playbook editor.

2. Write the prompt

A good playbook isn't just a raw prompt. It's a structured document your agent can follow reliably. Using the "Meeting Summary" example, here's what to fill in:

• Instructions — the core directive. E.g. "Read the meeting notes provided. Extract all action items with owners and due dates. Summarise key decisions in 3–5 bullet points. Write in a clear, professional tone."
• Inputs — what the user must provide when running it. E.g. "Raw meeting notes (paste or attach)"
• Outputs — what the agent should produce. E.g. "A list of action items, each with owner and due date. A bullet-point summary of key decisions."
• Examples — optional, but powerful. A short example of good output helps your agent stay on track, especially for formatting-sensitive tasks.

You don't need to get it perfect on the first try. Playbooks are versioned — you can refine them over time without losing what worked before.

3. Add context docs (optional)

You might notice that out of the box, your playbook's output sounds a bit generic. That's because your agent doesn't know anything specific about your team — your brand voice, your internal language, or how you like things formatted.

Context docs fix that. They're shared documents — things like a brand voice guide, a tone of voice brief, or a glossary of internal terms — that you attach to playbooks. Once attached, every run of that playbook has access to that context automatically.

If your meeting summaries feel impersonal, attaching a "Communication Style" context doc will make them sound like your team wrote them.

You can add context docs later — they don't block you from publishing. But if you have existing docs you'd like to add, see Getting Started for a walkthrough.

4. Approve your playbook

New playbooks start as drafts. A draft is visible to your team, but it signals "not ready yet."

When you're happy with the playbook, click Approve. This moves it to the approved state — the signal to your team that this is the version to use.

When you publish a new version later, Patina automatically deprecates the old one. Your team always sees a clean list of approved, current playbooks — no digging through drafts or guessing which version is right.

Approval answers the question: "How does my team know which is the good one?" The answer is: it has a green approved badge.

5. Use it from your AI tool

Once your playbook is approved, your connected AI tools can use it. In Claude Code, you might type:

Use the Meeting Summary playbook to summarise these notes: [paste notes here]

Your agent will fetch the playbook and any attached context from Patina, then follow the instructions you wrote — consistently, every time.

If you haven't connected your AI tool yet, head to Install the Plugin for setup instructions.