AI Prompt Notebook
Save, tag, and organize AI prompts for ChatGPT, Claude, Gemini, and more. Stored in your browser’s localStorage — nothing uploaded. Export as JSON, Markdown, or plain text.
Add Prompt
How to use
- Add a prompt: click “Add prompt,” fill in the title, paste your prompt body, pick a model target, add tags. Click Save.
- Search and filter: type in the search box to filter by title, body, tags, or notes. Use the model dropdown to narrow to a specific AI.
- Copy to clipboard: click Copy on any card to copy the prompt body — paste directly into ChatGPT, Claude, or Gemini.
- Back up regularly: click Export → JSON and save the file somewhere safe. Re-import any time via the Import button.
Frequently asked questions
Does this prompt notebook save to the cloud?
No. Everything is saved to your browser’s localStorage — a storage area on your device that is never transmitted to any server. Open DevTools → Application → Local Storage and you’ll see your prompts stored right there. Use Export → JSON to back up before clearing browser data.
What AI models are supported?
Any text-based model. You tag each prompt with a model target (ChatGPT, Claude, Gemini, Llama, or Other) to filter your library. The notebook doesn’t connect to any AI API — it’s purely an organizer for prompts you write yourself.
How do I export my prompts?
Click Export and choose JSON, Markdown, or Plain Text. JSON is best for backup and re-import. Markdown produces a document with headers per prompt, suitable for a notes app or GitHub Gist. Plain text lists title and body in sequence for easy pasting.
Can I import prompts from a backup?
Yes. Click Import and select a JSON file previously exported from this tool. Import merges by ID — no duplicates if you import the same file twice. If a prompt ID already exists locally, the imported version overwrites it.
How do tags work?
Enter tags separated by commas or spaces in the Tags field. They display as chips on each card. Click any tag chip to filter by it. Tags are stored in lowercase and match across title, body, and notes in the search box.
Is there a limit on how many prompts I can store?
Browser localStorage allows 5–10 MB per origin. A typical prompt is 0.5–2 KB, so you can store 2,000–10,000 prompts comfortably. If you hit the limit the tool will warn you and suggest exporting a backup before adding more.
What happens if I clear browser storage?
Your prompts will be deleted — there’s no server backup. Export a JSON file regularly and store it in Google Drive, Dropbox, or your desktop. The JSON is readable with any text editor and importable directly back into this tool.
How do I search for a specific prompt?
The search box filters by title, body, tags, and notes simultaneously as you type — no Enter needed. Combine it with the model filter dropdown to narrow results to a specific AI model.
Can I copy a prompt to my clipboard?
Yes. Each prompt card has a Copy button that sends the prompt body to your clipboard in one click, ready to paste into any AI. Use the Edit view to read the full body without the 3-line preview truncation.
Can I reorder or pin prompts?
Prompts sort by last-updated date (newest first). You can effectively pin a prompt by editing it — the timestamp updates and it sorts to the top. Prefix the title with ⭐ and search “⭐” to quickly find starred prompts.
Examples
Meeting summarizer (Claude)
Reusable prompt for turning raw transcripts into structured summaries with action items.
Title: Meeting summarizer Model: Claude Tags: meetings, productivity You are an executive assistant. Summarize the following meeting transcript. Output: 1. Key decisions (bullet list) 2. Action items: [owner] — [task] 3. Parking lot items Keep each section under 150 words.
Code reviewer (ChatGPT)
Drop in a function, get a structured review of correctness, edge cases, and naming.
Title: Code reviewer Model: ChatGPT Tags: dev, code-review Review the code below. For each issue, output: - Severity: low/medium/high - Line reference - Issue description - Suggested fix Focus on: correctness, edge cases, readability, security. Do not rewrite unless asked.
Email rewriter (Gemini)
Transform a rough draft into a polished professional email while preserving intent.
Title: Email rewriter Model: Gemini Tags: writing, email Rewrite the draft email below to be professional, concise, friendly. Preserve all key facts. Remove filler words. Keep under 150 words. Output only the rewritten email, no commentary.
About the AI Prompt Notebook
Every developer and power user who works with AI models accumulates dozens of carefully refined prompts. The problem is where to keep them. Notes apps lack model tagging. Browser bookmarks can’t store the prompt body. Cloud-based prompt managers require an account and upload your prompts to a third-party server — often one you haven’t read the privacy policy of.
This notebook stores everything in localStorage, the same mechanism your browser uses for app preferences and offline data. localStorage is strictly scoped to the site’s origin (prompts.platotools.com). No other website can read it, and nothing from it is ever transmitted to a server. Verify this: open DevTools → Network, use the tool, observe zero outbound requests. Open DevTools → Application → Local Storage and you’ll see your data stored right there as a JSON array under the key prompt_notebook_v1.
The privacy claim is literally verifiable because the tool has no backend. There is no database, no API endpoint, no server-side analytics on your prompt content. The only third-party scripts on this page are the Cloudflare Web Analytics beacon (which counts page views, not content) and Google AdSense (which serves ads). Neither has access to your localStorage.
The data model is intentionally simple. Each prompt stores: a unique ID, title, body (the actual instruction text), model target (ChatGPT, Claude, Gemini, Llama, or Other), optional tags (an array of lowercase strings), optional notes, and created/updated timestamps. The export function serializes this array directly to JSON — the file is human-readable and editable in any text editor.
Import merges by ID. When you import a backup, prompts with IDs that already exist locally are overwritten by the incoming version; new IDs are appended. This makes it safe to import the same file repeatedly without duplicates, and lets you share a curated prompt library as a JSON file that colleagues can import without overwriting their own prompts.
Browser localStorage typically provides 5–10 MB per origin. At an average of 1 KB per prompt (title, body, tags, notes combined), you can store thousands of prompts before approaching any limit. If you reach the quota, the tool surfaces a warning and blocks the save — prompting you to export a backup and free up space by deleting old prompts. The storage indicator at the top of the list gives you a live view of how much space is in use.
The search function scans title, body, tags, and notes simultaneously as you type. The model filter dropdown narrows the list to prompts tagged for a specific AI, making it fast to find the right prompt when you switch between ChatGPT and Claude in the same session. The Copy button on each card sends the body directly to your clipboard in one click.