🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.
TypeScript analyticsautogenevaluationlangchainlarge-language-modelsllama-indexllmllm-evaluationllm-observabilityllmopsmonitoringobservabilityopen-sourceopenaiplaygroundprompt-engineeringprompt-managementself-hostedycombinator
import {
SCORE_CONFIG_NAME_MAX_LENGTH,
SCORE_CONFIG_NAME_MIN_LENGTH,
} from "@langfuse/shared";
import { z } from "zod";
// Deliberately not reusing ScoreConfigNameSchema: its Unicode-aware regex is
// not representable in MCP JSON Schema, character rules are
// still enforced at runtime via the strict input schemas.
const McpScoreConfigNameSchema = z
.string()
.min(SCORE_CONFIG_NAME_MIN_LENGTH)
.max(SCORE_CONFIG_NAME_MAX_LENGTH)
.describe(
"Allowed characters: letters, numbers, spaces, underscores, periods, parentheses, and hyphens.",
);
const McpScoreConfigNumericMinValueSchema = z
.number()
.optional()
.describe("Minimum allowed value for NUMERIC score configs.");
const McpScoreConfigNumericMaxValueSchema = z
.number()
.optional()
.describe("Maximum allowed value for NUMERIC score configs.");
const McpScoreConfigCategoricalCategoriesSchema = z
.any()
.optional()
.describe(
"Allowed categories for CATEGORICAL score configs as an array of { label, value } objects.",
);
const normalizeMcpScoreConfigInput = <
TInput extends {
numericMinValue?: unknown;
numericMaxValue?: unknown;
categoricalCategories?: unknown;
},
>(
input: TInput,
) => {
const {
numericMinValue,
numericMaxValue,
categoricalCategories,
...scoreConfig
} = input;
return Object.fromEntries(
Object.entries({
...scoreConfig,
minValue: numericMinValue,
maxValue: numericMaxValue,
categories: categoricalCategories,
}).filter(([, value]) => value !== undefined),
);
};
const preprocessMcpScoreConfigInput = (input: unknown) => {
if (typeof input !== "object" || input === null || Array.isArray(input)) {
return input;
}
return normalizeMcpScoreConfigInput(input);
};
export {
McpScoreConfigCategoricalCategoriesSchema,
McpScoreConfigNameSchema,
McpScoreConfigNumericMaxValueSchema,
McpScoreConfigNumericMinValueSchema,
preprocessMcpScoreConfigInput,
};