🪢 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 { CommentObjectType } from "@langfuse/shared";
import { z } from "zod";
import { createCommentForApi } from "@/src/features/comments/server/publicCommentService";
import {
PostCommentsV1Body,
PostCommentsV1Response,
} from "@/src/features/public-api/types/comments";
import { defineTool } from "../../../core/define-tool";
import { buildCommentObjectUrl } from "@langfuse/shared/src/server";
import { runMcpTool } from "../../../core/run-mcp-tool";
const CreateCommentToolBaseSchema = z
.object({
content: z.string().trim().min(1).max(5000),
objectId: z.string(),
objectType: z.enum(CommentObjectType),
authorUserId: z.string().optional(),
})
.strict();
const CreateCommentToolSchema = PostCommentsV1Body.omit({
projectId: true,
});
export const [createCommentTool, handleCreateComment] = defineTool({
name: "createComment",
description: "Create a comment on a trace, observation, session, or prompt.",
action: "comments:CUD",
baseSchema: CreateCommentToolBaseSchema,
inputSchema: CreateCommentToolSchema,
handler: async (input, context) =>
runMcpTool({
spanName: "mcp.comments.create",
context,
attributes: {
"mcp.comment_object_type": input.objectType,
"mcp.comment_object_id": input.objectId,
},
fn: async () => {
const body = PostCommentsV1Body.parse({
...input,
projectId: context.projectId,
});
const result = await createCommentForApi({
input: body,
auditScope: context,
});
const comment = PostCommentsV1Response.parse(result);
const url = buildCommentObjectUrl({
projectId: context.projectId,
objectType: input.objectType,
objectId: input.objectId,
});
return url ? { ...comment, url } : comment;
},
}),
});