A browser-based compliment generator is a small writing helper that selects one line from a fixed inventory of 24 original prompts, where those prompts are arranged in 12 focus-and-tone pools so the wording stays relevant to the situation you picked rather than mixing a workplace acknowledgment with an unrelated playful observation. The Compliment Generator holds 4 focus values (character, creativity, kindness, and work) crossed with 3 tones (concise, playful, and warm), which produces the 12-pool structure (4 × 3 = 12), and every pool contains exactly 2 prompts, giving the full 24 lines (12 × 2 = 24). Selection runs through the browser's Web Crypto API rather than a server, name validation and formatting happen in the same tab, and nothing is uploaded, so the workflow stays private end to end. Because the inventory is fixed and the pool is small, the tool behaves more like a curated picker than a generative model: it chooses a line, applies the optional name prefix you provided, and hands you a plain-text sentence you can edit before sending. Knowing how those pieces fit together is what the rest of this article walks through.
The 24-Prompt Inventory and the 12-Pool Layout
The generator does not pull from a corpus or call a language model. It works from 24 original prompt records, each with a unique ID and text, written and reviewed as product copy. Those 24 records are not arranged in one big list. They are grouped into 12 pools by combining 4 focus values with 3 tone values, and every pool holds exactly 2 records. The arithmetic is straightforward:
- 4 focus values × 3 tone values = 12 focus-and-tone combinations (pools)
- 12 pools × 2 prompts per pool = 24 unique prompts total
Tests verify that all 24 IDs and texts are unique and that each of the twelve pools contains exactly two lines, so the inventory contract is fixed at the prompt level and does not depend on chance or external data. Every cell in the table below represents one pool of 2 lines.
| Focus \ Tone | Concise | Playful | Warm |
|---|---|---|---|
| Character | 2 lines | 2 lines | 2 lines |
| Creativity | 2 lines | 2 lines | 2 lines |
| Kindness | 2 lines | 2 lines | 2 lines |
| Work | 2 lines | 2 lines | 2 lines |
What Each Focus Controls
The focus picker decides the subject of the line, not how it sounds. None of the four values comment on appearance, health, intelligence, identity, or worth, so they stay inside the territory a sender can reasonably observe about someone they already know.
| Focus | What the lines emphasize |
|---|---|
| Character | Integrity, courage, honesty, dependability |
| Creativity | Ideas, imagination, new perspectives |
| Kindness | Attention, support, helping others feel seen |
| Work | Clarity, care, follow-through, contribution to a team |
A character line, for instance, names something like integrity or courage; a kindness line names attention or support. Because the focus controls the subject, switching focus without changing tone produces a line about a different aspect of the same person rather than a restatement of the original. This is also why a workplace acknowledgment stays distinct from a friendly check-in: the work pool cannot draw a line about a playful observation, and the kindness pool cannot draw a line about follow-through on a project.
What Each Tone Controls
Tone changes the delivery, not the underlying respect. The three values produce three distinct registers without rewriting the focus.
| Tone | Delivery style |
|---|---|
| Concise | Direct, single-sentence wording |
| Playful | Light imagery, slightly looser phrasing |
| Warm | Fuller sentence with extra context |
A concise character line and a warm character line both talk about integrity, but the concise version fits a chat message and the warm version fits a longer note or card. The tone value also decides which pool of 2 prompts is eligible for the next draw, so changing tone refreshes both the subject treatment and the candidate list in one step.
How Random Selection and the No-Repeat Guard Work
Selection runs in your browser tab. The tool calls crypto.getRandomValues to draw an unsigned 32-bit integer, then maps that value to one of the two prompts in the current focus-and-tone pool. Because the pool size does not divide 2³² evenly in general, the generator uses rejection sampling: it computes floor(2³² / poolLength) × poolLength as an upper bound and retries any value at or above that bound, capping retries at 128 to avoid a runaway loop. That step removes the modulo bias you would get if you took the random integer mod poolLength directly.
The second piece is the no-repeat guard. If a previous line was drawn from the same focus and tone, that prompt's ID is removed from the next pool before the random draw. With 2 prompts per pool, this means the next click in the same pool returns the other line instead of repeating the current one. Changing the focus, the tone, or the name field clears the old result, so the next draw starts from a fresh two-line pool.
Worked example for the current pool size of 2:
- Pool length = 2 prompts
- Upper bound = floor(2³² / 2) × 2 = 2,147,483,648 × 2 = 4,294,967,296, which equals 2³², so every 32-bit value is accepted and no rejection occurs.
- For a hypothetical pool of 3 prompts: upper bound = floor(2³² / 3) × 3 = 1,431,655,765 × 3 = 4,294,967,295, and because the maximum 32-bit value is also 4,294,967,295, only that single value is rejected, at a rate of about 1 in 4,294,967,296 (roughly 1 in 2³²) rather than 1 of every 3 values.
The contract is tested for valid boundaries, biased tail rejection, invalid-source failure, and no-repeat behavior, but those tests verify the selection mechanism. They do not prove any specific sentence fits every recipient or culture, which is why the sender still has the final say.
Name Handling and the Output Format
Personalization is optional. If you leave the name field empty, the generator returns the prompt text unchanged. If you enter a name, the widget trims whitespace at the outer edges, enforces a cap of 60 Unicode code points, and rejects control characters and Unicode line or paragraph separators rather than silently shortening over-limit names. Internal punctuation, accents, non-Latin writing, and emoji are preserved as entered, so the same rules cover names like María-José, 李雷, or Anya.
When a valid name is present, it is placed before the sentence and the original opening Your becomes lowercase your after the comma. Any line that opens with the capitalized word "Your" has that opening lowercased to "your" after the comma once a name is added, so a sentence that begins "Your ..." becomes "Maya, your ..." when a name is provided. The output stays plain text. The tool does not add an @mention, email address, greeting, signature, emoji, or quotation marks without your instruction, so what you see on screen is what gets put on the clipboard. Clipboard access happens only when you press the copy button, and a browser may deny that permission, in which case the wording stays visible for manual selection.
How to Use the Compliment Generator
- Optionally enter a name in the name field, keeping it within 60 Unicode code points and free of control characters; leave the field blank if a name would feel forced in the context.
- Choose the aspect you want to appreciate from the focus picker (character, creativity, kindness, or work) and pick a tone (concise, playful, or warm) that fits the relationship you have with the recipient.
- Press the generate control to draw a line from the two-prompt pool for that focus-and-tone combination; press it again to swap to the other prompt in the same pool without an immediate repeat.
- Read the line out loud and confirm it describes something genuinely observed, fits the relationship, and avoids comments on appearance, health, intelligence, identity, or worth.
- Personalize the wording by replacing any generic phrase with a concrete example (the careful review, patient explanation, thoughtful question, or creative solution that prompted the message), then press copy to put it on your clipboard.
The widget does not know the recipient, inspect a profile, analyze messages, infer personality, or decide whether a statement is true, so steps 4 and 5 are the only places where the sender's judgment actually enters the output. A full walkthrough of the same controls is in the Compliment Generator cheat sheet, and the live tool itself is at the Compliment Generator.
Privacy, Limits, and Where Editing Still Matters
Everything in this tool runs in the current browser tab. Name validation, cryptographic random selection, formatting, and the copy button all execute locally, and the widget does not submit the name, the focus, the tone, or the generated sentence to a server. Clipboard access occurs only when the copy button is pressed and may be denied by browser permissions, in which case the wording stays visible for manual selection. There is no game history, dice-style ledger, or analytics event tied to your prompt.
The limits worth remembering are these: 24 prompts total, 2 per focus-and-tone pool, 60 Unicode code points for an optional name, no decoration added to the output, and no automatic evaluation of the recipient. The lines are writing suggestions, not evidence-based judgments, and the inventory is product-authored creative copy rather than an external reference dataset. A short, truthful, context-aware message is usually stronger than an elaborate sentence, especially in professional, educational, or sensitive relationships where a generic line could read as romantic, patronizing, overly familiar, or based on a protected trait. Edit freely, add the concrete reason for your appreciation, and send only wording that sounds like you.
For a deeper look, see Can Labels Contain Commas When Using Line Graph Maker.