Enforces rigorous evidence-based validation before any task is marked as finished or successful.
The Verification Before Completion skill implements a mandatory 'Iron Law' for AI-driven development, preventing premature success claims by requiring fresh, empirical evidence. It transforms the AI workflow from an 'assume-done' approach to a 'verify-first' standard, ensuring that every assertion of completion—whether for bug fixes, builds, or feature development—is backed by real-time command output and error-free logs. This skill is essential for maintaining code integrity and preventing the common pitfall of AI agents reporting success based on partial or unverified assumptions.
主な機能
01Anti-hallucination guardrails for status reporting
02Automated verification for agent-delegated tasks
03Mandatory command execution before success claims
