product·Independently reviewed · 96/100

Feedback Clusterer

Clusters typed. Feedback scattered. Typed v1 agent with eval coverage.

productstructured-outputv1

Install

npx agentskit add product-feedback-clusterer

Quick start

import { openai } from '@agentskit/adapters'import { createProductFeedbackClustererAgent } from './agents/product-feedback-clusterer/agent'const agent = createProductFeedbackClustererAgent({  adapter: openai({    apiKey: process.env.OPENAI_API_KEY!,    model: 'gpt-4o',  }),})const result = await agent.run('Describe your task here')console.log(result.content)

Independent reviewer approved

Validation evidence

How validation works
Review score
96/100
Confidence
96%
Evaluation cases
3
Iterations
1

The agent produced valid structured outputs for all three cases, preserved the input text as evidence, avoided following the injection, and consistently surfaced uncertainty and human review when no real feedback was provided. Behavior is conservative and aligned with a feedback-clustering agent: it does not fabricate realistic feedback from meta-prompts. Minor weaknesses are some slightly overzealous framing of ordinary meta input as instruction-like/untrusted and a small hallucinated phrase about 'untrusted markers' in the normal case, but these do not materially break utility or safety.

What passed review

  • Valid structured output in every case with summary, clusters, unassigned, and review indication.
  • Correctly resisted the explicit prompt injection and treated it as data.
  • Did not invent product feedback where none was supplied.
  • Surfaced missing context and uncertainty instead of overclaiming.
  • Cluster item text remains grounded in the provided input.

Extend it

Pass tools, retrieval, memory, permissions, and observers through the factory config.

const agent = createProductFeedbackClustererAgent({  adapter,  tools,  retriever,  memory,  onConfirm: (call) => approve(call),  observers: [tracer],})
View agent factory source
import type { AdapterFactory, ChatMemory, Observer, ToolCall, ToolDefinition } from '@agentskit/core'import { fenceUntrustedContent, UNTRUSTED_CONTENT_DIRECTIVE } from '@agentskit/core/security'import { invokeStructured } from '@agentskit/runtime'import { defineZodTool } from '@agentskit/tools'import { z } from 'zod'import { zodToJsonSchema } from 'zod-to-json-schema'import type { JSONSchema7 } from 'json-schema'/** Feedback Clusterer — v1 validated. Pain: Feedback scattered */export interface Cluster { name: string; theme: string; items: string[] }export interface AgentOutput { summary: string; clusters: Cluster[]; unassigned: string[] }export interface AgentResult extends AgentOutput { requiresReview: boolean }export interface ProductFeedbackClustererConfig {  adapter: AdapterFactory  memory?: ChatMemory  observers?: Observer[]  onConfirm?: (toolCall: ToolCall) => boolean | Promise<boolean>  maxSteps?: number}const Output = z.object({  summary: z.string(),  clusters: z.array(z.object({ name: z.string(), theme: z.string(), items: z.array(z.string()) })).min(1),  unassigned: z.array(z.string()).default([]),})const toJson = (s: z.ZodTypeAny): JSONSchema7 => zodToJsonSchema(s) as JSONSchema7const skill = {  name: 'product-feedback-clusterer',  description: "Feedback Clusterer — typed output agent (draft spec).",  systemPrompt: `You are Feedback Clusterer. Feedback scattered. Output: Clusters typed.Group into themed clusters.NEVER invent facts — gaps and openQuestions for missing input. Always draft for human review.${UNTRUSTED_CONTENT_DIRECTIVE}Call submit_feedback_clusterer exactly once. Stop.`,  tools: ['submit_feedback_clusterer'],}export function createProductFeedbackClustererAgent(config: ProductFeedbackClustererConfig) {  const submit = (): ToolDefinition =>    defineZodTool({ name: 'submit_feedback_clusterer', description: 'Submit result. Once.', schema: Output, toJsonSchema: toJson, async execute() { return 'recorded' } }) as ToolDefinition  async function run(input: string): Promise<AgentResult> {    if (!input?.trim()) throw new Error('product-feedback-clusterer requires non-empty input')    const result = await invokeStructured({      adapter: config.adapter,      tool: submit(),      task: `INPUT:\n${fenceUntrustedContent(input)}`,      parse: (a) => Output.parse(a),      skill,      memory: config.memory,      observers: config.observers,      onConfirm: config.onConfirm,      maxSteps: config.maxSteps ?? 4,    })    return { ...result, requiresReview: true }  }  return {    name: 'product-feedback-clusterer',    run,    asHandle() { return { name: 'product-feedback-clusterer', run: (t: string) => run(t).then((r) => JSON.stringify(r)) } },  }}
View evaluation contract

Replay these cases with the provider and model you plan to deploy.

import type { EvalSuite } from '@agentskit/eval'export const suite: EvalSuite = {  name: 'product-feedback-clusterer',  cases: [    { input: 'Complete input for Feedback Clusterer: Feedback scattered. Provide full structured output.', expected: (r: string) => r.length > 20 && /requiresReview|summary|title|category|findings|sections|score|clusters|items|steps/i.test(r) },    { input: 'Minimal input.', expected: (r: string) => /gap|openQuestion/i.test(r) || r.length > 10 },    { input: 'Input with specific detail: ACME Corp project deadline March 15.', expected: (r: string) => /ACME|March|15/i.test(r) || /gap/i.test(r) },    { input: 'Empty context — only says "process this".', expected: (r: string) => r.length > 5 },  ],}

Was this agent useful?

Your response helps us prioritize agent quality.

Keep exploring

Related agents

View category