AI Output Reviewer
Paste your instructions on the left, the AI's output on the right. Hit review — the tool sends both to the AI (using your key) and gets back a structured assessment.
Key saved in your browser only. Sent to our server over HTTPS for the API call, then immediately discarded. Never stored or logged.
paste your instructions and AI output, then hit review
the AI will compare them and give you a structured assessment
Pattern matching can't catch logic errors
A regex can tell you that a variable is undefined. It can't tell you the code has a race condition, silently drops errors, or addresses 4 of your 5 requirements. That kind of assessment needs AI-level reasoning — which is what this tool provides.
What the review checks
- Compliance — Did the AI address every requirement in your instructions?
- Drift — What did it add that you didn't ask for? (Not all drift is bad.)
- Risks — Security issues, hardcoded values, removed error handling, edge cases.
- Confidence — Where the AI was hedging, guessing, or seemed uncertain.
- Follow-ups — Specific prompts to close the gaps the review identified.
Bring Your Own Key
This tool requires an API key from OpenAI, Anthropic, Google, or xAI. Your key is saved in your browser's localStorage (your device, your control). When you click review, the key is sent to our server over HTTPS, used for a single API call, and immediately discarded. It is never written to disk, database, or logs.
Cost: Each review uses your API tokens. A typical review with a few hundred words of instructions and a page of AI output costs roughly $0.01–0.05 depending on your provider and model. Token counts are shown after each review.