// docs

How to actually use it.

Task-oriented guides in reading order: follow 01 → 13 for the full tour, or jump straight to what you need. No pitch: commands, behavior, and the why behind each workflow.

Start here

01Getting startedAccount, first prompt, CLI install and login. Productive in five minutes.02Workspace & navigation⌘K palette, sidebar modes, the repositories index, repo home, and the editor prompt strip.

Core workflow

03Versioning, diff & rollbackCommit changes, compare any two versions, undo a bad edit, share a version-pinned link.04Templates with {{variables}}Turn a prompt into a fill-in template and use it from the web, the CLI, and your agents.05Pattern libraryProven prompt structures (CoT, few-shot, structured output) to start from instead of a blank page.06Token costsThe live ~token counter in the editor, your monthly AI budget, and MicroBoost top-ups.

AI tools

07AI ImproveproOne-shot variants of your prompt: compare, pick a winner, and it lands as a new version.08Loop: test with your own agentproGenerate variants, run each one through your real agent command, save the winner.

Terminal & agents

09CLI referenceEvery ple command with signatures and examples — works with no account since 0.4.0 (ple save, local Markdown store, explicit sync).10MCP server for agentsLet Claude Code (or any MCP client) read, search, and fill your prompts directly.

Share & collaborate

11Sharing & discoveryPublish a repo, claim your handle, get found in Explore, star, follow, and tag releases.12CollaborationInvite people with a link, viewer / editor / admin roles, and shared version history.

Your data

13Export & importTake your whole library out as JSON on any plan, or bulk-import existing prompts.

Questions that aren't workflows live in the Help & FAQ. Worked examples with real prompts are on the blog.