Agent Regression Test Prompt
Make sure a fix didn't break three other things — compare an agent's outputs across two versions on the same scenarios and flag every behavior that changed for the worse.
Prompt Utilities
Paste two texts and see exactly what changed — a git-style colored diff with additions in green, deletions in red, and word-level edits highlighted inline. It shows the mechanical difference and nothing else: no score, no verdict, no "which is better." For what a prompt revision means there's the Prompt Version Diff; for which prompt wins there's the Prompt Comparator.
Make sure a fix didn't break three other things — compare an agent's outputs across two versions on the same scenarios and flag every behavior that changed for the worse.
Run a prompt twice, or on two models, and diff the answers to see exactly where they differ — mechanically, without ranking them.
Two versions of a document, compared line by line — which sections were added, removed, or reordered, at a glance.
Generated two variants of a headline or a line of copy? Diff them to see precisely which words changed between the drafts.
Drop two pieces of text side by side and get the literal difference between them — what was added, removed, and reworded.
Did anything change, and what exactly? Detect every change between two versions — a swapped number, an added word — with nothing missed.
The changes you would miss on a re-read — an added word, a swapped term — surfaced exactly, so nothing slips past.
Compare two blocks line by line — config, lists, or records — to see exactly which lines were added, removed, or changed.
A free, browser-based diff tool: paste two texts, get a colored diff and a change count. Nothing installed, nothing uploaded.
Paste two texts and see exactly what changed — a git-style colored diff with additions, deletions, and word-level edits, and a mechanical change count.
When the edits are inside a sentence, line mode is too coarse. Word mode highlights the exact words inserted and deleted.
Restructure code you own without breaking it — change only what's worth changing, and prove with tests and a diff that behavior held.
Find out whether an AI agent behaves before users do — define what correct means, build test scenarios with expected outputs, catch failures and hallucinations, then regression-test each version.
Make any AI task return JSON your code can rely on — define the schema, force the model to it, validate every response, and diff the drift when a model update breaks the shape.
Paste the original text in the first box and the changed text in the second, then pick a diff mode. The Diff Engine compares them with an LCS algorithm: Line mode treats each line as a unit (added, removed, or unchanged); Word mode highlights the exact words inserted and deleted inside changed regions; Mixed mode (the default) does both — line-level structure with word-level highlighting inside similar line pairs. Click Compare Texts for a git-style colored diff: additions in green, deletions in red, word-level edits marked inline, with a mechanical Change Summary — N lines added, removed, changed, and N words inserted and deleted. The copyable report breaks it into Added, Removed, and Modified content plus a unified diff and statistics. Everything runs in your browser, and all text is HTML-escaped before display, so the diff is safe. The tool reports only what changed — it does not score quality, recommend, assess risk, or say which text is better.
Mechanics versus meaning. The Prompt Version Diff interprets two versions of a prompt: it analyzes facet changes (role, tone, format), risk deltas, and delivers a verdict — "Version B is a net improvement" or "a regression" or "mostly cosmetic". This tool shows the literal difference between any two texts — the words that were added, removed, and changed — and stops there. One tells you what a prompt revision means; the other shows you exactly what moved.
Showing versus deciding. The Prompt Comparator scores two independent prompts across quality dimensions and tells you which one is better and why. This tool neither scores nor recommends; it displays the difference between two texts and leaves every judgment to you. Same verb — "compare" — opposite intent: the Comparator decides a winner, the Diff Checker just shows the change.
Any text. It is a general-purpose mechanical diff — documents, AI outputs, config files, contract clauses, lists, code snippets, drafts. There is no assumption that the input is a prompt, which is exactly what separates it from the prompt-specific tools. If you can paste it, it can diff it.
Line mode compares whole lines and reports each as added, removed, or unchanged — best for config, lists, and structured text. Word mode highlights the exact words inserted and deleted inside changed regions — best for reworded prose where the line break does not matter. Mixed mode, the default, combines them: it keeps line structure but, for similar removed-and-added line pairs, highlights which words changed inside the line. Most edits read best in Mixed.
No, by design. The Change Summary is a purely mechanical tally — lines and words added, removed, and changed. The tool offers no quality score, no recommendation, no risk assessment, and no "which is better". Its whole job is to show the difference accurately and let you decide what it means. If it judged the change, it would be a different tool.
It is safe. Because a diff has to render the texts you paste, every piece of your content is HTML-escaped before it is shown — a script tag or angle brackets in your input appear as literal characters in the diff, never as live markup. The colored view is built only from escaped content plus the tool's own markup, so pasting code, HTML, or markup is fine.
Yes — the same two texts always produce the same diff, because it's a deterministic algorithm, not an AI comparison. Copy or download the diff report yourself.