Skip to main content
Back to challenges
Challenge

Automated Content Moderation Workflow with OpenAI Agents SDK

In response to concerns about AI moderation errors, develop a robust multi-agent team that analyzes user content submissions. Utilizing the OpenAI Agents SDK, you will orchestrate a specialist team where one agent acts as a content screener, another performs sentiment validation, and the final agent produces human-readable explanations. This system uses Hume AI to detect emotional context in audio-based social content, ensuring moderation decisions are nuanced and auditable.

Workflow AutomationHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Develop an autonomous moderation agent swarm using OpenAI Agents SDK to improve transparency in content governance.

Delivery guide

How work is evaluated

Evaluation

Moderation accuracy audit

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

ModerationLogicTest

Check if agent adheres to safety guidelines

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

ModerationThroughput

Requests per minute • target: 50 • range: 10-100

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Master the OpenAI Agents SDK pattern for defining multi-turn autonomous teams

  • Integrate Hume AI for context-aware moderation of social audio inputs

  • Automate documentation and log auditing for moderation events using Aider

  • Design agentic workflows in Twin to connect moderation tools with enterprise APIs

  • Evaluate moderation accuracy by implementing a human-in-the-loop verification step

  • Scale agent decision making using GPT-5.4-mini for efficient reasoning under load

Resources and assets

Reference links and supporting material

Dataset notes

Sample data for 1 tasks

Challenge setup

Moderation accuracy audit

Challenge input

Voice recording clip

OpenAI

OpenAI AI model provider

Hume AI

Emotionally intelligent voice AI

Twin

AI automation platform

Evaluated output

Boolean flag and explanation

Evaluation checks
  • Check if agent adheres to safety guidelines
  • Requests per minute • target: 50 • range: 10-100
Configured evaluation evidence
  • ModerationThroughput target: 50
  • 1 public reference case
Configured runtime
  • JavaScript execution harness

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • OpenAI · Required
  • Hume AI · Optional
  • Twin · Optional

Evaluation contract

  • ModerationLogicTest · Weight 1
  • ModerationThroughput · Weight 1

Metrics

  • ModerationThroughput · Target 50

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: Voice recording clip

    Output: Boolean flag and explanation

    Execution harness: JavaScript

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Automated Content Moderation Workflow with OpenAI Agents SDK