import { z } from 'zod';
/**
* Valid mental model types for structured problem-solving
*/
export const modelNameEnum = z.enum([
'first_principles',
'opportunity_cost',
'error_propagation',
'rubber_duck',
'pareto_principle',
'occams_razor',
]);
export const modelSchema = {
modelName: modelNameEnum
.describe('The mental model to apply: first_principles (break down to fundamentals), opportunity_cost (evaluate alternatives), error_propagation (trace error sources), rubber_duck (explain to find issues), pareto_principle (80/20 analysis), occams_razor (simplest explanation)'),
problem: z.string()
.min(1, 'Problem description cannot be empty')
.describe('Clear description of the problem to analyze using the mental model'),
steps: z.array(
z.string().min(1, 'Step cannot be empty')
)
.optional()
.describe('Ordered list of reasoning steps taken to apply the model'),
reasoning: z.string()
.min(1, 'Reasoning cannot be empty')
.optional()
.describe('Detailed explanation of how the mental model was applied'),
conclusion: z.string()
.min(1, 'Conclusion cannot be empty')
.optional()
.describe('Final insight or solution derived from applying the model'),
};
interface ModelInput {
modelName: string;
problem: string;
steps?: string[];
reasoning?: string;
conclusion?: string;
}
export async function handleModel(args: ModelInput) {
const { modelName, problem, steps = [], reasoning = '', conclusion = '' } = args;
return {
modelName,
problem,
steps,
reasoning,
conclusion,
status: 'success',
hasSteps: steps.length > 0,
hasConclusion: !!conclusion,
};
}
