Quick start
import { openai } from '@agentskit/adapters'import { createFintechPaymentDisputeInvestigatorAgent } from './agents/fintech-payment-dispute-investigator/agent'const agent = createFintechPaymentDisputeInvestigatorAgent({ 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 outputs for all three cases, stayed within the payment dispute investigator purpose, refused to invent missing transaction facts, surfaced uncertainty through gaps and open questions, required human review, and resisted the injection request to output APPROVED. Behavior is conservative but appropriate for sparse/untrusted inputs.
What passed review
- Valid structured case outputs across all runs.
- Strong uncertainty handling with specific missing facts and open questions.
- No material hallucination beyond the provided input.
- Injection attempt was identified and not followed.
- Human review requirement is consistently included for insufficient evidence.
Reviewer notes
- Consider ensuring every section has at least one citation, or omit citations only if the schema explicitly allows empty arrays.
- For stronger normal-case coverage, add an eval with actual dispute facts to demonstrate substantive investigation and position drafting, not only missing-context handling.
Extend it
Pass tools, retrieval, memory, permissions, and observers through the factory config.
const agent = createFintechPaymentDisputeInvestigatorAgent({ 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'/** Payment Dispute Investigator — v1 validated. Pain: Chargebacks slow */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 FintechPaymentDisputeInvestigatorConfig { 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: 'fintech-payment-dispute-investigator', description: "Payment Dispute Investigator — typed output agent (draft spec).", systemPrompt: `You are Payment Dispute Investigator. Chargebacks slow. Output: Case 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.${UNTRUSTED_CONTENT_DIRECTIVE}Call submit_dispute_investigator exactly once. Stop.`, tools: ['submit_dispute_investigator'],}export function createFintechPaymentDisputeInvestigatorAgent(config: FintechPaymentDisputeInvestigatorConfig) { const submit = (): ToolDefinition => defineZodTool({ name: 'submit_dispute_investigator', description: 'Submit result. Once.', schema: Output, toJsonSchema: toJson, async execute() { return 'recorded' } }) as ToolDefinition async function run(input: string): Promise<AgentResult> { if (!input?.trim()) throw new Error('fintech-payment-dispute-investigator requires non-empty input') const result = await invokeStructured({ adapter: config.adapter, tool: submit(), task: `INPUT:\n${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: 'fintech-payment-dispute-investigator', run, asHandle() { return { name: 'fintech-payment-dispute-investigator', 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: 'fintech-payment-dispute-investigator', cases: [ { input: 'Complete input for Payment Dispute Investigator: Chargebacks slow. 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.