LiveReview

Description

LiveReview is an AI-powered code review tool from Hexmos that scores every pull request and merge request for risk, so engineering teams can spend human review time on the changes most likely to cause problems in production.

Its risk score blends semantic LLM analysis, a custom machine learning model trained on an organisation's own production incidents and reverts, and static graph-based analysis of the whole codebase. The approach is inspired by Meta's RADAR review system, and the lowest-risk 30 to 50 percent of changes can be auto-approved under a conservative, balanced or aggressive policy, while risky or uncertain changes still go to a person.

Key features include blast-radius scoring, review-priority ranking, a 10-category risk taxonomy covering more than 100 failure patterns, an Issue Navigator for filtering findings by severity and category, AI-generated pull request summaries, 60-second summary slide decks, and PR quizzes that check a developer understands a change before it is committed.

Reviews can run at several levels: on commit through the git-lrc hook, before push, on the pull request, in CI/CD and on a schedule. Teams can add repository rules in a .lrc directory, write jq-based CI/CD gates that block or allow a merge, and use Adaptive Reviews, which pairs a leader model with a cheaper helper model to cut inference costs by 40 to 50 percent.

For reporting, LiveReview offers a one-click Impact Report with 57 charts and Livi, an AI chatbot that answers questions about engineering data with charts.

It works with GitHub, GitLab, Bitbucket, Gitea and Azure DevOps, and is available on the web, as a CLI, via a Claude Code integration (claude-lrc), and as extensions for VS Code, Cursor and Antigravity. It also connects to Slack, Microsoft Teams and Discord, and offers an MCP server and REST API. Teams can use its fine-tuned model or bring their own key for Gemini, OpenAI, AWS Bedrock or Ollama.

Pricing: a free plan includes 30,000 reviewed lines of code per month. Paid plans start at $32 for 100,000 lines, scale by reviewed lines rather than seats, and include unlimited users. Self-hosted and air-gapped deployments, SSO and SLA support are available on the Enterprise plan.

It is suited to engineering teams and leaders who ship a growing volume of AI-written code and want consistent review standards, audit trails and fewer production incidents.

The codebase is source-available on GitHub.