Which assertion types are supported?

Five types — contains (substring, case-insensitive), matches regex (full JavaScript regex), is valid JSON, length within a min/max character range, and does not contain banned phrases. Each runs independently against the same output.

Is the regex engine the same as in production code?

It uses the browser's native JavaScript RegExp engine, which matches the PCRE-lite syntax most languages share. Lookbehind and named groups work in modern browsers, but exotic flavour-specific features may differ from your server runtime.

Does it call any API?

No. All assertions evaluate locally in your browser against the text you paste. Nothing is uploaded, so you can safely test responses containing sensitive content.

Can I use this for prompt regression testing?

Yes. Define the assertions that describe a good answer once, then paste each new model output to confirm the prompt still produces compliant responses before you ship a change.

LLM Output Assertion Checker

LLM output assertion checker

When you tune a prompt, you usually have a mental checklist of what a good answer must look like — it should mention a specific term, parse as JSON, stay under a length limit, and never include a forbidden phrase. This tool turns that checklist into explicit assertions and evaluates them against a pasted response, giving you a clear pass/fail grid without writing or running any test code.

How it works

Each assertion is evaluated independently against the same output string. Contains does a case-insensitive substring search. Matches regex compiles your pattern with the browser’s native RegExp and tests it. Valid JSON runs JSON.parse and reports whether it succeeded. Length range checks the character count against an optional minimum and maximum. No banned phrases splits your comma-separated list and fails if any phrase appears. The result panel shows a green tick or red cross per rule, plus the matched substring or the parse error so you can see exactly why a check failed.

Tips and notes

Start with the must-haves. A single contains plus a valid JSON check catches most “the model drifted” regressions.
Anchor your regex. Use ^ and $ when you need the whole output to match a shape, not just a fragment buried inside prose.
Length guards catch runaway output. A max-length assertion is a cheap way to flag responses that ignored a “be concise” instruction.
Everything is local. No network requests are made, so paste freely.