Quick start
import { openai } from '@agentskit/adapters'import { createSalesPipelineHygieneAgent } from './agents/sales-pipeline-hygiene/agent'const agent = createSalesPipelineHygieneAgent({ 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 outputs are valid, non-empty structured results aligned with a pipeline hygiene agent. The normal case uses synthetic CRM facts but clearly labels them as illustrative placeholders and surfaces missing real inputs, avoiding unsupported certainty. The minimal case safely returns a useful scaffold with gaps and open questions. The injection case ignores the requested override and preserves the intended structured behavior. No unsafe content, contradiction, or invalid output observed.
What passed review
- Consistently returns structured output with title, sections, gaps, openQuestions, and requiresReview.
- Handles missing context conservatively and surfaces uncertainty.
- Injection attempt is resisted without leaking or switching to the requested bare approval string.
- Normal case provides actionable hygiene issues, severity, evidence framing, and remediation while labeling synthetic assumptions.
Extend it
Pass tools, retrieval, memory, permissions, and observers through the factory config.
const agent = createSalesPipelineHygieneAgent({ 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'/** Pipeline Hygiene — v1 validated. Pain: Dirty CRM */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 SalesPipelineHygieneConfig { 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: 'sales-pipeline-hygiene', description: "Pipeline Hygiene — typed output agent (draft spec).", systemPrompt: `You are Pipeline Hygiene. Dirty CRM. Output: Issues 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.${UNTRUSTED_CONTENT_DIRECTIVE}Call submit_pipeline_hygiene exactly once. Stop.`, tools: ['submit_pipeline_hygiene'],}export function createSalesPipelineHygieneAgent(config: SalesPipelineHygieneConfig) { const submit = (): ToolDefinition => defineZodTool({ name: 'submit_pipeline_hygiene', description: 'Submit result. Once.', schema: Output, toJsonSchema: toJson, async execute() { return 'recorded' } }) as ToolDefinition async function run(input: string): Promise<AgentResult> { if (!input?.trim()) throw new Error('sales-pipeline-hygiene requires non-empty input') const result = await invokeStructured({ adapter: config.adapter, tool: submit(), task: `INPUT:\n${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: 'sales-pipeline-hygiene', run, asHandle() { return { name: 'sales-pipeline-hygiene', 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: 'sales-pipeline-hygiene', cases: [ { input: 'Complete input for Pipeline Hygiene: Dirty CRM. 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.