Applies to
reviewTool: 'coderabbit'only. The host review tool defaults toreviewTool: 'dakar'; this section describes the retained CodeRabbit mode.
With coderabbitHostReview enabled (the default within CodeRabbit mode), the
workflow host — not the task agents — runs
coderabbit review --agent --type committed against each task branch: once per
dual-review round (alongside the code and expert reviewers) and once per
addendum implementation. Because only committed changes are reviewed, the
ExecPlan durability contract doubles as the review contract. The host parses
the CLI's structured findings; critical and major severities join the
reviewers' blocking items and drive the ordinary fix rounds, while lower
severities are captured without gating integration.
Rate limits are absorbed by the host: a rate-limited review waits a
deterministic 45–90 minutes (hostReviewBackoffMinutes) and retries, up to
hostReviewAttempts total attempts, costing wall-clock but zero agent tokens.
A rate limit that outlives every attempt — or a CLI fault — defers the review
with a documented openIssues entry on the task result instead of blocking
integration; the dual reviewers remain decisive. A CodeRabbit authentication
failure halts the task as fatal-auth.
The run result's tool-neutral hostReview object reports the selected
reviewer, effective configuration, and bounded counters for runs, findings,
retries, deferred outcomes, timeouts, errors, authentication failures, and sink
failures. Terminal logs record the bounded reviewer and review label, terminal
attempt count, elapsed milliseconds, outcome, and error category; host-side
timeouts are classified from process metadata. When hostReviewFindingsFile is
set, every finding is appended through the serialized JSONL sink. Sink failures
are reported through hostReview and do not fail the task. The historical
CodeRabbit type and runner names remain compatibility aliases only.