orc-analyze

What problem does it solve? Requirements documents and plain-language requests often conflict with what the code actually does, contain unstated assumptions, or bleed into scopes the user never asked for. This Skill turns a requirement (a document, PDF, or bare request) into a confirmed, code-grounded requirement set before any planning, so implementation never builds against stale claims or unverified assumptions. ## Core Features & Use Cases - Code-grounded reconciliation: Maps every requirement to real files with quote-anchored file:line evidence, classifying each as exists, missing, conflict, or buildable; unverifiable claims become explicit clarifying questions. - Scope bounding with anchored context: Bounds the deliverable to exactly the asked scope while pulling adjacent scopes in only as non-actionable, anchored "do not build" context. - Standard and Deep modes: Standard mode verifies a stated evidence floor; opt-in Deep mode adds orchestrator-dispatched parallel scouts, verify-every-claim coverage, and an alternatives-and-risks analysis. - Use Case: Given a security audit PDF, run "/orc-analyze security-audit.pdf — implement the AUTH findings only" to get a confirmed report.md and derived requirement-spec.md where stale findings are challenged and dropped before any build planning starts. ## Quick Start Ask the assistant to run /orc-analyze on your requirements document or plain-language feature request and confirm the detected mode and scope when prompted.