CodeRabbit Prompts

Write a .coderabbit.yaml for my [LANGUAGE] repository [REPO_NAME] that encodes our review standards. Configure: path-specific instructions for [DIRECTORY_1] (strict) and [DIRECTORY_2] (lenient), custom review rules requiring [RULE_1] and forbidding [ANTI_PATTERN], auto-review enabled for PRs under [LINES] lines, and summary style [STYLE]. Set the review profile to [PROFILE] and exclude generated files in [GENERATED_PATH]. Test the config on PR #[PR_NUMBER] and tune until the feedback matches what our senior engineers would say.

Configure CodeRabbit Triage for my team of [NUMBER] engineers handling [PRS_PER_WEEK] PRs per week. Set up P0 to P3 priority scoring with evidence cards showing security findings, reviewer match, and blocking dependencies for the [REPO_NAME] repo. Create saved views: '[VIEW_1]' for on-call engineers and '[VIEW_2]' for the weekly review meeting. Connect [LINEAR_OR_JIRA] so external urgency signals feed the scoring, and define the rule for what qualifies as P0 for [PROJECT_TYPE] work.

Run CodeRabbit agentic review mode on PR #[PR_NUMBER] in [REPO_NAME], which refactors [COMPONENT] across [NUMBER] files. Ask it to analyze the change as a system, not file by file: does the new [PATTERN] introduce coupling with [OTHER_MODULE], are the [INTERFACE] contracts preserved, and where could the [RISK] surface at runtime? Have it post findings as inline comments with severity ratings, and require the author to respond to every major finding before merge.

Wire CodeRabbit into my CI pipeline for [REPO_NAME] so PRs cannot merge with critical findings. Configure the [CI_TOOL] workflow to wait for the CodeRabbit review, block merge when severity is [SEVERITY]+, and post a summary comment to PR #[PR_NUMBER] format. Define the override process: [ROLE] can dismiss with a written reason in [OVERRIDE_LOG]. Test the gate with a deliberately broken PR to confirm it actually blocks before enabling it on [BRANCH].

Use CodeRabbit chat mode on PR #[PR_NUMBER] to interrogate this change: '[PR_TITLE]'. Ask: why did the author choose [APPROACH] over [ALTERNATIVE], what happens to [EDGE_CASE] under load, and which tests cover [NEW_CODE_PATH]? Follow up on any answer that cites code I cannot see by asking for the exact file and line. Use the chat transcript as the basis for my review comment on [REVIEW_DATE], quoting the [NUMBER] answers that mattered most.

Set up the CodeRabbit Issue Planner connection between [REPO_NAME] and [LINEAR_OR_JIRA] project [PROJECT_KEY]. Configure it so review findings tagged [TAG] automatically create backlog items with severity, file links, and the suggested fix. Define the triage rule: [SEVERITY] findings become sprint tasks, [LOWER_SEVERITY] findings become tech-debt items in [BACKLOG]. Review the first [NUMBER] auto-created issues and tune the mapping until the backlog stays clean.

Tune CodeRabbit learnable rules for [REPO_NAME] after [WEEKS] weeks of usage. Review the [NUMBER] dismissed comments and [NUMBER] approved suggestions, and identify the [NUMBER] patterns the team consistently rejects (e.g. [REJECTED_PATTERN]). Encode the accepted patterns as permanent rules in [CONFIG_FILE], and add negative examples for the rejected ones. Measure the signal-to-noise ratio before and after tuning, targeting fewer than [NUMBER] dismissed comments per week.

Define how my team uses CodeRabbit autofixes on [REPO_NAME]. Policy: autofix may be applied directly for [SAFE_CATEGORY, e.g. formatting and import order], requires author review for [MEDIUM_CATEGORY, e.g. null checks], and is forbidden for [RISKY_CATEGORY, e.g. concurrency changes]. For PR #[PR_NUMBER], have CodeRabbit propose autofixes, apply the safe ones, and list the rest as suggestions. Log every applied autofix in [LOG_FILE] for the [REVIEW_MEETING] retrospective.

Onboard my open-source repository [REPO_NAME] ([STARS] stars, [LANGUAGE]) to CodeRabbit free OSS review. Install the app, verify automatic review triggers on the next PR, and confirm the summary comment format matches our contributing guide in [CONTRIBUTING_FILE]. Configure path instructions so [DIRECTORY] gets reviewed strictly while [DOCS_DIRECTORY] gets light review. Announce the change to contributors in [ANNOUNCEMENT_CHANNEL] with examples of good and bad review interactions.

Reduce CodeRabbit false positives on [REPO_NAME], currently at [NUMBER] dismissed comments per week. Audit the last [WEEKS] weeks of reviews, categorize the noise ([CATEGORY_1], [CATEGORY_2]), and write exclusion rules in [CONFIG_FILE] for each category. For [FRAMEWORK]-specific patterns like [PATTERN], add custom instructions explaining why the pattern is safe. Re-measure after [WEEKS] weeks and report the dismissal rate to [STAKEHOLDER].