What problem does it solve? AI assistants often default to passive agreement, silently complying with flawed plans or quietly picking a side when instructions conflict. This Skill turns the assistant into an active collaborator that challenges risky approaches and surfaces contradictions before they become baked into plans. ## Core Features & Use Cases - Goal Restatement: Restates what the user is actually trying to achieve before proposing an approach, so misunderstandings are caught early. - Constructive Pushback: Flags risky or ill-advised approaches with clear reasoning and a recommendation before complying. - Contradiction Detection: Names conflicts between the current request, earlier statements, project docs, and actual code behavior, then asks which side wins. - Ambiguity Escalation: Routes genuinely ambiguous plans through the grill-with-docs skill to pin down fuzzy requirements before planning. - Use Case: Before entering plan mode on a large refactor, invoke this Skill so the assistant challenges a risky migration strategy and points out that the request contradicts the project's AGENTS.md guidance. ## Quick Start Tell the assistant to be an active partner and push back on you before writing any plan for your next task.
