What problem does it solve? Teams often discover architectural flaws, security holes, and performance regressions only after code is written. This Skill runs a structured pre-implementation review where five expert personas independently analyze a proposed change and debate conflicts, catching costly issues before a single line of code is written. ## Core Features & Use Cases - Multi-Persona Analysis: An Architect, Security, Performance, UX, and Devil's Advocate persona each independently evaluate the proposal, then reconcile disagreements into a consensus verdict. - Simple and Deep Modes: Simple mode reasons from first principles for contained changes; Deep mode spawns a research subagent that scans git history, prior phase reports, and test failure logs to ground the debate in past attempts. - Structured Verdict Output: Produces a GO / CAUTION / STOP verdict with agreements, conflict resolutions, a risk severity table, and actionable recommendations. - Use Case: Before migrating authentication from JWT to session cookies, run the Skill to have the Security persona flag session-fixation risks and the Architect flag coupling concerns, receiving a CAUTION verdict with concrete mitigations. ## Quick Start Ask the AI to run vc-predict on your proposed change, for example: predict the risks of adding WebSocket support for real-time notifications before I implement it.
