0.5.2

CLI v0.5.2

v0.5.2 improves output visibility, adds two commands for inspecting past reviews, and makes agent-mode pipelines more reliable.

Plain-text output now includes a finding count and per-severity breakdown at the end of each run; individual finding cards are easier to scan mid-output.
coderabbit review findings replays locally cached results from the last run. coderabbit review --show-prompts prints the AI prompts used in that run.
Agent mode now emits heartbeat events during long reviews, includes a comment field as a fallback when codegenInstructions is absent, and warns when plain mode is used inside a detected agent environment.