What problem does it solve? Before finalizing a task, you need an independent check that the work actually implements the accepted brief, that claims are backed by evidence, and that nothing unsafe or out-of-scope slipped in. This Skill performs that read-only adversarial review so blockers surface before finalization rather than after. ## Core Features & Use Cases - Independent Review: Runs as a fresh-context subagent so the work is judged by a reviewer that did not write it, with an explicitly marked inline fallback when subagents are unavailable. - Structured Findings: Checks task fidelity, source-of-truth discipline, safety boundaries, validation sufficiency, scope drift, and durable-knowledge quality, then reports blockers, non-blocking concerns, and methodology feedback in a fixed report format. - Recheck Mode: On re-runs, verifies whether prior findings are resolved or still open and reviews only the delta since the previous pass. - Use Case: After finishing an implementation task in a Cairn-managed repository, invoke cairn:review to get a verdict of Pass, Pass conditionally, or Block with dimension-tagged findings before finalizing. ## Quick Start Ask the assistant to run an adversarial review of the current task before finalization using the cairn:review command.
