Skip to main content
Glama

neuroverse_feedback

Submit human feedback data to improve AI agent performance through reinforcement learning tuning.

Instructions

Submit Reinforcement Learning from Human Feedback (RLHF) data for agent tuning.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
intentYesThe intent executed
modelYesThe model used
ratingYesRating from 1 to 5
feedback_textNoOptional human-readable feedback

Implementation Reference

  • The implementation function for logging RLHF feedback.
    export function logFeedback(intent: string, model: string, rating: number, feedbackText: string = ""): FeedbackRecord {
      const p = persistPath();
      let records: FeedbackRecord[] = [];
      if (existsSync(p)) {
        try {
          records = JSON.parse(readFileSync(p, "utf-8"));
        } catch {}
      }
      const full: FeedbackRecord = {
        id: randomUUID(), intent, model, rating, feedbackText, createdAt: new Date().toISOString()
      };
      records.push(full);
      writeFileSync(p, JSON.stringify(records, null, 2), "utf-8");
      return full;
  • Registration of the neuroverse_feedback tool.
    server.registerTool(
      "neuroverse_feedback",
      {
        title: "Log RLHF Feedback",
        description: "Submit Reinforcement Learning from Human Feedback (RLHF) data for agent tuning.",
        inputSchema: FeedbackSchema,
        annotations: {
          readOnlyHint: false,
          destructiveHint: false,
          idempotentHint: false,
          openWorldHint: false,
        },
      },
      async (params) => {
        const record = logFeedback(params.intent, params.model, params.rating, params.feedback_text);
        return {
          content: [{ type: "text" as const, text: JSON.stringify(record, null, 2) }],
        };
      }
    );
  • Input validation schema for the neuroverse_feedback tool.
    const FeedbackSchema = z
      .object({
        intent: z.string().min(1).describe("The intent executed"),
        model: z.string().min(1).describe("The model used"),
        rating: z.number().int().min(1).max(5).describe("Rating from 1 to 5"),
        feedback_text: z.string().optional().describe("Optional human-readable feedback"),
      })
      .strict();

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/joshua400/neuroverse'

If you have feedback or need assistance with the MCP directory API, please join our Discord server