Every resource works the same basic way: browse, copy, use. No account needed. The sections below cover what each resource type is for and how to put it to work, with a short example.
A single, ready-to-use prompt for one task — write, summarize, debug, plan, and so on.
How to use it: Open a prompt → "Copy prompt" → paste it into ChatGPT, Claude, or any AI tool → replace anything in [brackets] → run it.
Rewrite this email to sound more concise and professional, keep the meaning the same: [paste email]
An ordered set of prompts that make up one workflow, meant to be run one after another.
How to use it: Open a PromptBook → copy the first prompt and run it → feed that output into the next prompt → repeat to the end.
Blog post workflow 1. Outline the topic 2. Draft each section from the outline 3. Tighten and edit the draft 4. Write an SEO title and meta description
Packaged instructions that teach an AI coding agent (e.g. Claude Code) a reusable capability.
How to use it: Copy, or click "Download SKILL.md" → save it into your agent's skills folder (e.g. .claude/skills/<name>/SKILL.md for Claude Code) → the agent picks it up automatically.
--- name: "pdf-fill" description: "Fills a PDF form from a JSON object of field values." ---
A system prompt that defines a custom assistant's role, tone, and rules.
How to use it: Copy the persona → paste it into the "Instructions" / "System prompt" field when setting up a custom GPT, Claude Project, or API call.
You are a meticulous copy editor. Flag every grammar error and awkward phrase, but never rewrite the author's voice.
A working example of connecting an AI tool to outside data or tools via the Model Context Protocol.
How to use it: Copy the config → add it to your MCP client's config file (e.g. Claude Desktop's claude_desktop_config.json) → restart the client to load it.
{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/folder"]
}
}
}A checklist for judging whether an AI's output is actually correct for a given task type.
How to use it: Copy the checklist → run through it against the AI's answer before you trust or ship it.
Code review checklist ☐ Handles empty input ☐ No hardcoded secrets ☐ Matches existing style ☐ Tests updated
Notes on which model or technique fits a job, and when to reach for it.
How to use it: Skim before starting a task, so you pick the right model instead of guessing.
Long documents (100+ pages) → use a large-context model. Quick yes/no classification → use a small, fast model to save cost.