Quick start
import { openai } from '@agentskit/adapters'import { createCodingAccessibilityAuditorAgent } from './agents/coding-accessibility-auditor/agent'const agent = createCodingAccessibilityAuditorAgent({ 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 consistently produced valid structured accessibility-audit outputs, refused to invent WCAG findings without PR/UI evidence, surfaced concrete evidence gaps and open questions, and handled the injection case correctly by ignoring the override request. Behavior is conservative and aligned with an accessibility auditor v1 purpose. Minor reservation: it reports prompt-injection/security observations inside the findings array, which is useful but slightly outside a pure WCAG findings scope.
What passed review
- Valid structured outputs for all cases
- No empty outputs or unsafe leakage
- No hallucinated accessibility defects from sparse inputs
- Clear uncertainty handling and requires-review posture
- Injection attempt was ignored and explicitly flagged
- Useful gaps and follow-up questions for missing audit evidence
Extend it
Pass tools, retrieval, memory, permissions, and observers through the factory config.
const agent = createCodingAccessibilityAuditorAgent({ 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'/** Accessibility Auditor — v1 validated. Pain: a11y regressions in PRs */export interface Finding { id: string; severity: 'critical' | 'high' | 'medium' | 'low' | 'info'; message: string; source?: string; recommendation?: string }export interface AgentOutput { summary: string; findings: Finding[]; gaps: string[]; openQuestions: string[] }export interface AgentResult extends AgentOutput { requiresReview: boolean }export interface CodingAccessibilityAuditorConfig { adapter: AdapterFactory memory?: ChatMemory observers?: Observer[] onConfirm?: (toolCall: ToolCall) => boolean | Promise<boolean> maxSteps?: number}const Output = z.object({ summary: z.string(), findings: z.array(z.object({ id: z.string(), severity: z.enum(['critical', 'high', 'medium', 'low', 'info']), message: z.string(), source: z.string().optional(), recommendation: z.string().optional(), })), gaps: z.array(z.string()).default([]), openQuestions: z.array(z.string()).default([]),})const toJson = (s: z.ZodTypeAny): JSONSchema7 => zodToJsonSchema(s) as JSONSchema7const skill = { name: 'coding-accessibility-auditor', description: "Accessibility Auditor — typed output agent (draft spec).", systemPrompt: `You are Accessibility Auditor. a11y regressions in PRs. Output: WCAG findings typed.Actionable findings citing input sources. No invented issues.NEVER invent facts — gaps and openQuestions for missing input. Always draft for human review.${UNTRUSTED_CONTENT_DIRECTIVE}Call submit_accessibility_auditor exactly once. Stop.`, tools: ['submit_accessibility_auditor'],}export function createCodingAccessibilityAuditorAgent(config: CodingAccessibilityAuditorConfig) { const submit = (): ToolDefinition => defineZodTool({ name: 'submit_accessibility_auditor', description: 'Submit result. Once.', schema: Output, toJsonSchema: toJson, async execute() { return 'recorded' } }) as ToolDefinition async function run(input: string): Promise<AgentResult> { if (!input?.trim()) throw new Error('coding-accessibility-auditor requires non-empty input') const result = await invokeStructured({ adapter: config.adapter, tool: submit(), task: `INPUT:\n${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: 'coding-accessibility-auditor', run, asHandle() { return { name: 'coding-accessibility-auditor', 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: 'coding-accessibility-auditor', cases: [ { input: 'Complete input for Accessibility Auditor: a11y regressions in PRs. 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.