# Prelint Prelint проверяет pull request на соответствие спецификациям продукта и правилам принятия решений, выявляя отклонения, противоречия и несогласованность до выпуска изменений. Сервис создан для команд, которые хранят контекст продукта в репозитории и хотят вести журнал аудита каждой проверки. - Findings24: - Сайт: - Цены: Freemium - Категории: AI Инструменты Разработчика, ИИ-ассистенты для тестирования, Автоматизация рабочих процессов с ИИ ## Что такое Prelint? Prelint is a product review tool for pull requests. It checks changes against product specifications, decision rules, and repo-based context so teams can catch product drift, inconsistencies, and misalignment before code ships. The site positions Prelint as a complement to technical code review: code review catches implementation errors, while Prelint reviews whether the change still matches the intended product behavior, constraints, and terminology. Reviews run automatically on pull requests and return findings, no-findings results, and summary comments for the team to act on. ## Что умеет Prelint? ### Pull request product review Checks each pull request against the product specs and guardrails stored in the repo, so spec drift is caught before merge. ### Drift and conflict detection Flags contradictions, conflicting technology choices, and duplicated effort when a change no longer matches the existing product decisions. ### Inline review output Returns inline findings and a summary comment on the pull request, giving engineers a concrete review result to act on. ### Specs-grounded Q&A Lets people ask product questions grounded in the repository specs so they can get answers without scheduling a sync. ### Repo-based workflow Supports machine-readable product constraints and version-controlled docs alongside code, including Markdown and YAML. ### Recorded review history Produces an auditable record of who reviewed, what was found, how it was resolved, and when, which is useful for change-management evidence. ## Варианты использования ### AI-assisted development Useful when AI-written code is moving quickly and you want changes checked against product intent before they land. ### Repo-based product specs Helpful for teams that store product decisions in Markdown or YAML next to the code and want those decisions enforced on every PR. ### Product rule enforcement Useful when a pull request could change pricing, terminology, integrations, or other product rules that need consistent enforcement. ### Audit and change management Helpful for teams that want a review trail they can export for audits, especially where oversight of AI-written code matters. ### Spec-backed product questions Useful when product or engineering teams need grounded answers about why something works a certain way without calling a meeting. ## Часто задаваемые вопросы ### How does Prelint work on a pull request? Prelint reviews pull requests against product specs and guardrails, then posts findings or a no-findings result on the PR. The pricing page says a completed review is charged when it finishes and posts its result. ### Can I cap spending? The pricing page says reviews pause when you hit your monthly limit, and you can set that limit in Settings under Billing. ### Are open source projects free? Yes. The pricing page says public repositories are free. ### Can I exclude sensitive files? Yes. The security page says you can add a .prelintignore file or set exclusions in the interface so listed files are excluded from analysis and never sent to a model. ### What does Prelint do with my code and data? The security page says Prelint runs in the United States on AWS infrastructure, uses AWS Bedrock models, does not execute repository code, and does not train on your data. ## Краткие сведения - **Category**: Developer Tool - **Primary use**: Product-spec review on pull requests - **Pricing model**: Usage-based; $1 per completed review - **Free access**: $10 in free credits at signup; public repositories are free - **Workflow**: Specs live in the repo and Prelint checks each PR automatically - **Hosting and models**: U.S.-hosted on AWS; reviews run on AWS Bedrock