Terms of service
Last updated: November 7, 2026.
The service
DiffReview provides AI-generated code reviews of diffs you paste and of public GitHub pull requests you link. One full review per day is free. Additional reviews are sold as one-time point packs: $9 for 20 reviews, $29 for 100 reviews. Points never expire and are tied to your account.
The review is advice, not a guarantee
The review is generated by an AI model. It can miss real issues, and it can flag things that are not problems. It does not guarantee that your code is correct, complete, secure, or free of bugs. Always have a human review your code before shipping. You are responsible for deciding what to do with the review's suggestions. We are not liable for bugs, security incidents, or other harm in code you ship after using the service.
What you may submit
Submit only code you have the right to share: your own code, or code you are authorized to review. Do not submit code that belongs to someone else without permission, and do not submit secrets, credentials, or private keys in a diff — the review does not need them, and anything you paste is sent to the AI model once for the review (then deleted). Do not paste private repository content unless you have the right to share it; note that a pasted diff is treated as content you chose to share.
Only public GitHub pull requests can be fetched by URL. The service has no GitHub login and cannot access private repositories.
Fair use
Reviews are limited to 30,000 characters / 1,000 lines per diff, 10 review requests per IP per hour, and a shared daily AI budget. We may rate-limit or block abuse (for example, automated bulk submission) without notice.
Accounts and payments
You are responsible for keeping your account credentials safe. Purchases are final except as described in the refund policy. We may change prices for new purchases; existing points keep their value.
Availability
The service runs on a shared free AI quota; if the daily budget is exhausted, reviews pause until the next day. We aim for high availability but provide the service "as is", without uptime guarantees.