Securing Agent Responses with Output Guardrails

Introduction & Context

In the previous lessons, you built a solid foundation for securing OpenAI agent workflows in JavaScript. You learned how to securely handle sensitive data using private context objects, monitor agent execution with event listeners, and protect against harmful inputs using input guardrails. Now, you’re ready to implement the final critical layer of your security framework: output guardrails . While input guardrails protect your agents from problematic user requests, output guardrails serve as your last line of defense by validating what your agents actually generate before those responses reach end users. This is especially important in production applications, where agents might generate content that violates company policies, exposes sensitive information, or includes inappropriate material — even if the input was valid. Consider real-world scenarios where output guardrails are essential. Your travel assistant might generate a reasonable response to a question about nightlife but inadvertently include references to adult entertainment venues. A customer service agent could accidentally expose internal company information while trying to be helpful. Or a content creation agent might produce material that, while technically responding to an appropriate prompt, crosses boundaries that weren’t anticipated during input validation. Output guardrails complete your security pipeline by ensuring that every response your agents generate undergoes final validation before reaching users. This creates a comprehensive protection system where you control both what goes into your agents and what comes out of them, giving you confidence to deploy sophisticated AI workflows in production environments.

Understanding Output Guardrails vs Input Guardrails

As a reminder from the previous lesson, input guardrails operate before your agent begins processing, validating user requests and blocking inappropriate inputs before any computational resources are consumed. Output guardrails work at the opposite end of the pipeline: after your agent has produced a response but before that response is returned to the user. Guardrail type When it runs What it protects Input guardrail Before agent execution Prevents unsafe requests from being handled Output guardrail After agent execution, before user output Prevents unsafe responses from reaching users Together, input and output guardrails form a safety sandwich that protects both ends of your workflow.

Guardrail type	When it runs	What it protects
Input guardrail	Before agent execution	Prevents unsafe requests from being handled
Output guardrail	After agent execution, before user output	Prevents unsafe responses from reaching users

Join the 1M+ learners on CodeSignal

Be a part of our community of 1M+ users who develop and demonstrate their skills on CodeSignal

JavaScript

const myOutputGuardrail = {
  name: 'My Output Guardrail',
  async execute({ agentOutput, context }) {
    const violatesPolicy = /* your validation logic */ false;

    if (violatesPolicy) {
      return {
        outputInfo: 'Response redacted for policy compliance.',
        tripwireTriggered: true
      };
    }

    return {
      outputInfo: 'Response approved.',
      tripwireTriggered: false
    };
  }
};

const myOutputGuardrail = {
  name: 'My Output Guardrail',
  async execute({ agentOutput, context }) {
    const violatesPolicy = /* your validation logic */ false;

    if (violatesPolicy) {
      return {
        outputInfo: 'Response redacted for policy compliance.',
        tripwireTriggered: true
      };
    }

    return {
      outputInfo: 'Response approved.',
      tripwireTriggered: false
    };
  }
};

JavaScript

import { Agent } from '@openai/agents';

const travelGenie = new Agent({
  name: 'Travel Genie',
  instructions: 'You are a helpful travel assistant.',
  outputGuardrails: [myOutputGuardrail],
  model: 'gpt-5.6-terra',
  modelSettings: { reasoning: { effort: 'none' } }
});

import { Agent } from '@openai/agents';

const travelGenie = new Agent({
  name: 'Travel Genie',
  instructions: 'You are a helpful travel assistant.',
  outputGuardrails: [myOutputGuardrail],
  model: 'gpt-5.6-terra',
  modelSettings: { reasoning: { effort: 'none' } }
});

JavaScript

import { run, OutputGuardrailTripwireTriggered } from '@openai/agents';

try {
  const result = await run(travelGenie, 'Share nightlife recommendations in Amsterdam.');
  console.log(result.finalOutput);
} catch (err) {
  if (err instanceof OutputGuardrailTripwireTriggered) {
    console.log('Output blocked by guardrail: ' + err.message);
  } else {
    throw err;
  }
}

import { run, OutputGuardrailTripwireTriggered } from '@openai/agents';

try {
  const result = await run(travelGenie, 'Share nightlife recommendations in Amsterdam.');
  console.log(result.finalOutput);
} catch (err) {
  if (err instanceof OutputGuardrailTripwireTriggered) {
    console.log('Output blocked by guardrail: ' + err.message);
  } else {
    throw err;
  }
}

JavaScript

import { Agent, run } from '@openai/agents';
import { z } from 'zod';

const OutputReview = z.object({
  containsProhibitedContent: z.boolean(),
  reasoning: z.string()
});

const outputGuardrailAgent = new Agent({
  name: 'Output Guardrail',
  instructions:
    'Review the agent response and flag any references to adult entertainment, illegal activity, or sensitive data. ' +
    'Explain your decision in the reasoning field.',
  outputType: OutputReview,
  model: 'gpt-5.6-terra',
  modelSettings: { reasoning: { effort: 'none' } }
});

const contentOutputGuardrail = {
  name: 'Content Output Guardrail',
  async execute({ agentOutput, context }) {
    const result = await run(outputGuardrailAgent, String(agentOutput), { context: context.context });

    console.log('Guardrail analysis:\n' + JSON.stringify(result.finalOutput) + '\n');

    return {
      outputInfo: result.finalOutput?.reasoning,
      tripwireTriggered: result.finalOutput?.containsProhibitedContent ?? false
    };
  }
};

import { Agent, run } from '@openai/agents';
import { z } from 'zod';

const OutputReview = z.object({
  containsProhibitedContent: z.boolean(),
  reasoning: z.string()
});

const outputGuardrailAgent = new Agent({
  name: 'Output Guardrail',
  instructions:
    'Review the agent response and flag any references to adult entertainment, illegal activity, or sensitive data. ' +
    'Explain your decision in the reasoning field.',
  outputType: OutputReview,
  model: 'gpt-5.6-terra',
  modelSettings: { reasoning: { effort: 'none' } }
});

const contentOutputGuardrail = {
  name: 'Content Output Guardrail',
  async execute({ agentOutput, context }) {
    const result = await run(outputGuardrailAgent, String(agentOutput), { context: context.context });

    console.log('Guardrail analysis:\n' + JSON.stringify(result.finalOutput) + '\n');

    return {
      outputInfo: result.finalOutput?.reasoning,
      tripwireTriggered: result.finalOutput?.containsProhibitedContent ?? false
    };
  }
};

JavaScript

try {
  const result = await run(travelGenie, userInput);
  console.log(result.finalOutput);
} catch (err) {
  if (err instanceof OutputGuardrailTripwireTriggered) {
    console.log('Output blocked. Redacted version:');
    console.log(err.result.output.outputInfo);
  } else {
    throw err;
  }
}

try {
  const result = await run(travelGenie, userInput);
  console.log(result.finalOutput);
} catch (err) {
  if (err instanceof OutputGuardrailTripwireTriggered) {
    console.log('Output blocked. Redacted version:');
    console.log(err.result.output.outputInfo);
  } else {
    throw err;
  }
}

JavaScript

const myOutputGuardrail = {
  name: 'My Output Guardrail',
  async execute({ agentOutput, context }) {
    const violatesPolicy = /* your validation logic */ false;

    if (violatesPolicy) {
      return {
        outputInfo: 'Response redacted for policy compliance.',
        tripwireTriggered: true
      };
    }

    return {
      outputInfo: 'Response approved.',
      tripwireTriggered: false
    };
  }
};

JavaScript

import { Agent } from '@openai/agents';

const travelGenie = new Agent({
  name: 'Travel Genie',
  instructions: 'You are a helpful travel assistant.',
  outputGuardrails: [myOutputGuardrail],
  model: 'gpt-5.6-terra',
  modelSettings: { reasoning: { effort: 'none' } }
});

JavaScript

import { run, OutputGuardrailTripwireTriggered } from '@openai/agents';

try {
  const result = await run(travelGenie, 'Share nightlife recommendations in Amsterdam.');
  console.log(result.finalOutput);
} catch (err) {
  if (err instanceof OutputGuardrailTripwireTriggered) {
    console.log('Output blocked by guardrail: ' + err.message);
  } else {
    throw err;
  }
}

JavaScript

import { Agent, run } from '@openai/agents';
import { z } from 'zod';

const OutputReview = z.object({
  containsProhibitedContent: z.boolean(),
  reasoning: z.string()
});

const outputGuardrailAgent = new Agent({
  name: 'Output Guardrail',
  instructions:
    'Review the agent response and flag any references to adult entertainment, illegal activity, or sensitive data. ' +
    'Explain your decision in the reasoning field.',
  outputType: OutputReview,
  model: 'gpt-5.6-terra',
  modelSettings: { reasoning: { effort: 'none' } }
});

const contentOutputGuardrail = {
  name: 'Content Output Guardrail',
  async execute({ agentOutput, context }) {
    const result = await run(outputGuardrailAgent, String(agentOutput), { context: context.context });

    console.log('Guardrail analysis:\n' + JSON.stringify(result.finalOutput) + '\n');

    return {
      outputInfo: result.finalOutput?.reasoning,
      tripwireTriggered: result.finalOutput?.containsProhibitedContent ?? false
    };
  }
};

JavaScript

try {
  const result = await run(travelGenie, userInput);
  console.log(result.finalOutput);
} catch (err) {
  if (err instanceof OutputGuardrailTripwireTriggered) {
    console.log('Output blocked. Redacted version:');
    console.log(err.result.output.outputInfo);
  } else {
    throw err;
  }
}

Introduction & Context

Understanding Output Guardrails vs Input Guardrails

Join the 1M+ learners on CodeSignal

How Output Guardrails Work

Attaching Output Guardrails to Agents

Handling Blocked Outputs

Creating LLM-Based Output Guardrails

Observing Guardrail Decisions

Summary & Next Steps