/** * Semantic Gate Definitions * * Extends the base gate system with semantic analysis gates that detect * common anti-patterns in user prompts before workflow activation. * * New gate types: * - no_shortcut_keywords: detects minimising language ("just", "simply") * - has_verification_plan: requires test/verify intent in prompt * - proper_decomposition: ensures task scope is appropriate * - flag_semantic_check: validates flags match prompt complexity */ /** Gate result shared interface */ export interface SemanticGateResult { passed: boolean; gate: string; description: string; blocking: boolean; details?: string; } /** * Detect shortcut language that encourages shallow solutions. */ export declare function checkShortcutKeywords(prompt: string): SemanticGateResult; /** * Check if the prompt includes intent to verify or test the outcome. */ export declare function checkVerificationPlan(prompt: string): SemanticGateResult; /** * Ensure the task is properly decomposed for complex work. * Simple tasks (<80 chars, low complexity) always pass. * Complex tasks should have multiple steps or sub-tasks indicated. */ export declare function checkDecomposition(prompt: string): SemanticGateResult; /** * Validate that flags match the prompt's semantic complexity. * E.g., --deliberate on a trivial task is a mismatch. */ export declare function checkFlagSemantic(prompt: string, flags: string[], _skill: string): SemanticGateResult; /** * Run all semantic gates against a prompt. */ export declare function runSemanticGates(prompt: string, flags?: string[], skill?: string): SemanticGateResult[]; /** * Check if any semantic gate is blocking. */ export declare function hasBlockingGates(results: SemanticGateResult[]): boolean; /** * Format gate results as a human-readable message. */ export declare function formatGateResults(results: SemanticGateResult[]): string; //# sourceMappingURL=gate-definitions.d.ts.map