← Back to all sparks
C

CodeRabbit

DEVOPS
Velocity6.3

AI-powered code review platform that posts line-by-line pull request feedback

CodeRabbit moves its review status to a GitHub check run and keeps Deep Scan coverage fresh.

code-reviewsecurity-scanninggithubcontinuous-coverageci-checks
◆Current state
CodeRabbit is building a security product on top of its PR reviewer. October brought AI Deep Scan API access on all plans, linked-repo context, merge blocking on Security findings, runtime reproducers and now Delta Deep Scans that refresh changed or stale coverage between full scans. On the review side, the status now posts as a GitHub check run with per-stage timing.
◆Where it's heading
Deep Scan is turning from a periodic audit into continuous coverage wired into merge gates. The review product is integrating deeper into GitHub's native primitives and reading other agents' config, such as Claude Code rules.
◆Prediction
Expect Delta Deep Scans and runtime verification to extend beyond GitHub to GitLab, following the pattern of earlier triage and linking features.

◆Recent moves

  1. 18h ago

    IDE extension 0.25: split findings, Fix with AI hand-off

    The IDE extension catches up with the cloud reviewer: findings and nitpicks are split and grouped by file, and Fix with AI can hand either set to an agent. It brings the same review-then-hand-off loop CodeRabbit has built on GitHub into the editor.

    View source ↗
  2. 1d ago

    CLI 0.9.1: fail reviews by severity, manage agent skills

    The CLI can now fail a review at a chosen severity, which turns it into a usable CI or agent gate, and it reads the review base from .coderabbit.yaml. Skill management and credit reporting extend the agent-facing CLI line started in v0.9.0.

    View source ↗
  3. 1d ago

    Review check run | GitHub | CodeRabbit Cloud

    The review status on GitHub becomes a check run listing each review stage and its duration, with skipped and rate-limited reviews shown as neutral. It makes CodeRabbit look like a native CI step, matching its push into merge gates.

    View source ↗
  4. 2d ago

    Delta Deep Scans | Continuous coverage | CodeRabbit Cloud

    Delta Deep Scans refresh changed code and stale coverage between full scans, with automation for GitHub repos. It moves Deep Scan from a one-off audit toward continuous coverage, which the merge-blocking check needs to stay accurate.

    View source ↗
  5. 6d ago

    Deep Scan findings get reproducer guides and runtime checks

    Deep Scan findings now come with step-by-step reproducer guides, or can be run inside CodeRabbit with Runtime Analysis. It targets false positives, the main cost of AI security findings.

    View source ↗
  6. 9d ago

    Claude Code rules in Code Guidelines | PR Reviews

    CodeRabbit reads Claude Code rules from .claude/rules, honouring path scopes. Teams that already configured one agent get consistent reviews without duplicating guidelines.

    View source ↗