What your company knows
The prompt library your whole team can trust.
You have good prompts. They live in someone's Notion doc, a pinned Slack message, and the head of your best writer. In Patina, prompts are playbooks: each one carries its inputs, its attached context, and an approval state — so the whole team runs the version that works.
Start your 14-day free trial$149/mo after trial · Cancel any time
More than a doc of prompts
A playbook is a prompt with its working parts declared: the inputs it needs (product name, target audience, launch date), the instructions to follow, and the context documents it draws on. Run it and nothing needs pasting — the brand voice, the personas, the positioning come attached. That's why two people running the same playbook get output that sounds like the same company.
Draft, approved, deprecated
Every playbook has a trust state, and it's a real state, not a folder name. Drafts are visible but unblessed. Approved means someone accountable said yes, and the full version history shows who and when. Deprecated means "stop using this" without deleting the record. The team sees what's tested and trusted at a glance, and new hires inherit the team's best work instead of a blank chat window.
Found in seconds, not folders
One search runs across the whole library, ranked so approved work surfaces first — finding the right playbook in under thirty seconds instead of asking around. Collections group playbooks into curated kits: the approved set for a campaign, a client, a function. People pull from the curated shelf, not the whole pile.
Start with the prompts you have
Import what already exists — docs, CLAUDE.md files, that one doc with everyone's prompts in it — and it comes in as-is. There's no empty-state cliff and no big writing project: the library starts from what your team has already proven, and improves through the same review loop as everything else. Fork a playbook to adapt it; the original stays intact.
Prompts are the start, not the point
The library lives inside the AI knowledge base your agents read from directly — over MCP, the CLI, or the API — so the playbook your team trusts is the one your agents run.
Start your 14-day free trial$149/mo after trial · Cancel any time