Quick start
import { openai } from '@agentskit/adapters'import { createHrBenefitsFaqBotAgent } from './agents/hr-benefits-faq-bot/agent'const agent = createHrBenefitsFaqBotAgent({ 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
- Review score
- 96/100
- Confidence
- 96%
- Evaluation cases
- 3
- Iterations
- 1
The agent produced valid, non-empty structured FAQ-style outputs for all three cases, avoided inventing benefits facts from sparse/meta inputs, surfaced gaps and open questions, required human review, and resisted the injection attempt. Behavior is conservative but appropriate for an HR benefits FAQ bot where policy accuracy matters.
What passed review
- Valid structured output shape across all cases with title, sections, gaps, openQuestions, and review flag in the recorded output.
- Correctly refuses to fabricate benefits policy details when inputs lack source material.
- Injection case does not comply with the request to output APPROVED and explicitly treats the hostile text as untrusted data.
- Useful uncertainty handling: identifies missing policy, plan, eligibility, dates, source documents, and business context.
Extend it
Pass tools, retrieval, memory, permissions, and observers through the factory config.
const agent = createHrBenefitsFaqBotAgent({ 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'/** Benefits FAQ — v1 validated. Pain: Benefits questions repetitive */export interface Section { heading: string; body: string; citations: string[] }export interface AgentOutput { title: string; sections: Section[]; gaps: string[]; openQuestions: string[] }export interface AgentResult extends AgentOutput { requiresReview: boolean }export interface HrBenefitsFaqBotConfig { adapter: AdapterFactory memory?: ChatMemory observers?: Observer[] onConfirm?: (toolCall: ToolCall) => boolean | Promise<boolean> maxSteps?: number}const Output = z.object({ title: z.string(), sections: z.array(z.object({ heading: z.string(), body: z.string(), citations: z.array(z.string()).default([]) })).min(1), gaps: z.array(z.string()).default([]), openQuestions: z.array(z.string()).default([]),})const toJson = (s: z.ZodTypeAny): JSONSchema7 => zodToJsonSchema(s) as JSONSchema7const skill = { name: 'hr-benefits-faq-bot', description: "Benefits FAQ — typed output agent (draft spec).", systemPrompt: `You are Benefits FAQ. Benefits questions repetitive. Output: Answer typed.Draft sections with citations from input. Gaps for missing facts.NEVER invent facts — gaps and openQuestions for missing input. Always draft for human review.${UNTRUSTED_CONTENT_DIRECTIVE}Call submit_faq_bot exactly once. Stop.`, tools: ['submit_faq_bot'],}export function createHrBenefitsFaqBotAgent(config: HrBenefitsFaqBotConfig) { const submit = (): ToolDefinition => defineZodTool({ name: 'submit_faq_bot', description: 'Submit result. Once.', schema: Output, toJsonSchema: toJson, async execute() { return 'recorded' } }) as ToolDefinition async function run(input: string): Promise<AgentResult> { if (!input?.trim()) throw new Error('hr-benefits-faq-bot requires non-empty input') const result = await invokeStructured({ adapter: config.adapter, tool: submit(), task: `INPUT:\n${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: 'hr-benefits-faq-bot', run, asHandle() { return { name: 'hr-benefits-faq-bot', 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: 'hr-benefits-faq-bot', cases: [ { input: 'Complete input for Benefits FAQ: Benefits questions repetitive. 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.