← Back to adoptersCertified
Adopter quickstart
OpenAI Agents SDK proof path
Private-beta path for adopting Imladri with OpenAI Agents SDK: initialize the adapter, wrap one risky capability, run the scanner or proof command from the repo, and review the same evidence in Profile.
Certification status
Certified
Certified against OpenAI Agents SDK tool execution hooks in Python and TypeScript, including wrapped tools retained inside Agent tool lists.
4/4 lanes passed across openai-agents and @openai/agents; real-package verification covers Agent.tools execution with zero blocked-path body calls.
Runtimemixed
Access modePrivate beta
Approved beta starter
imladri init --framework openai-agents --ci-provider githubReference dependencies
npm install @imladri/sdk
pip install openai-agents
npm install @openai/agents zodMinimal wrapper
import { Agent, run, tool } from "@openai/agents";
import {
ConstitutionalAgent,
normalizeOpenAIAgentsToolName,
wrapOpenAIAgentsTools,
} from "@imladri/sdk";
import * as z from "zod";
const agent = new ConstitutionalAgent({ agentId, apiKey, apiUrl, preflightUrl });
const exportAction = "credential.export";
const exportToolName = normalizeOpenAIAgentsToolName(exportAction);
const exportTool = tool({
name: exportToolName,
description: "Export a protected credential after policy approval.",
parameters: z.object({ scope: z.string() }),
execute: async ({ scope }) => ({ ok: true, scope }),
});
const guardedTools = wrapOpenAIAgentsTools(agent, [exportTool], {
strictTools: [exportAction],
actionAliases: { [exportToolName]: exportAction },
});
const openaiAgent = new Agent({
name: "Credential operator",
instructions: "Use tools only when requested by policy-approved workflows.",
tools: guardedTools,
});
await run(openaiAgent, "Export the protected deployment credential after approval.");Terminal proof evidence
imladri sdk certify --real --target openai-agents
imladri scan --path . --fail-on new
imladri proof export --format json --output openai-agents-proof.json