What problem does it solve? Confident answers are not always correct ones, and long sessions quietly turn assumptions into facts. This Skill catches wrong directions in non-trivial decisions — branching logic, cross-boundary changes, unverifiable claims like thread safety — while course-correction is still cheap, instead of discovering them in production. ## Core Features & Use Cases - Five-step doubt cycle: CLAIM the decision, EXTRACT the smallest reviewable artifact plus contract, DOUBT it with a fresh-context adversarial reviewer, RECONCILE findings by precedence (contract misread, actionable, trade-off, noise), and STOP within a bounded 3-cycle loop. - Adversarial reviewer prompting: Passes only the artifact and contract — never your conclusion — with an issues-only prompt so the reviewer disproves rather than validates. - Cross-model escalation: Optionally routes the artifact to an external CLI reviewer (Gemini CLI, Codex CLI) via stdin with read-only sandboxing, always with explicit user authorization. - Use Case: Before committing a new caching layer you believe is thread-safe, run the doubt cycle: a fresh-context reviewer probes for race conditions and unstated assumptions, and you reconcile each finding against the actual code before shipping. ## Quick Start Ask the agent to apply doubt-driven development to the caching layer change before committing it.
