What problem does it solve? Writing software specifications from a single perspective often misses edge cases, unstated assumptions, and constraint interactions, leading to ambiguous implementation targets and costly rework. This Skill produces a structured NLSpec specification grounded in multi-AI research, adversarial review, and completeness validation. ## Core Features & Use Cases - Multi-AI Research Probe: Dispatches specification research through orchestrate.sh to external providers (Codex, Antigravity) and synthesizes the consensus findings. - Structured NLSpec Template: Generates a complete spec with Purpose, Actors, Behaviors (triggers, preconditions, steps, postconditions, edge cases), Constraints, Dependencies, and an Acceptance Definition with a satisfaction target. - Adversarial Completeness Challenge: Routes the draft spec to a different provider (or a Sonnet agent) to find missing requirements, overlooked constraints, and untestable postconditions before finalizing. - Use Case: Run /octo:spec user authentication system to answer clarifying questions about actors and constraints, then receive a validated spec.md with 3-7 testable behaviors and a satisfaction target calibrated to complexity. ## Quick Start Ask the assistant to create an NLSpec specification for your feature, for example: generate a structured specification for a user authentication system covering actors, behaviors, and constraints.
