Halberd
Inspiration
Halberd was born out of real needs we've encountered while moderating Discord servers for school events. As officers and presidents of clubs, we've often faced the challenge of harmful and spammy content disrupting our community spaces. Existing tools were either too rigid or too limited, so we wanted to build a more intelligent, flexible solution. That's where Halberd comes in-- a fully configurable AI-powered moderation agent, designed to act swiftly and intelligently based on user-specified rules.
What it does
Halberd automatically detects and responds to harmful content according to customizable moderation rules. It supports multimodal inputs-- meaning it can analyze not just text but images too. Users can define complex, highly specific rulesets to detect almost any kind of content, and Halberd will enforce them automatically.
It also features an advanced data analytics dashboard to help users monitor moderation performance, fine-tune rulesets, and optimize behavior for different communities.
How we built it
- Frontend: Built using React for a responsive, intuitive user interface.
- Backend: Powered by FastAPI for robust and scalable server-side logic.
- Bot Logic: Integrated with Discord using the official Discord Python SDK.
- Authentication: Implemented Google OAuth 2.0 for secure user login.
- AI: Incorporated OpenAI models for content understanding and moderation.
Challenges we ran into
One major challenge was integrating with unfamiliar technology stacks, especially when setting up the Discord bot and ensuring it behaved correctly across a variety of server environments. Deployment was another learning curve — choosing the right cloud services and configuring everything properly took time and experimentation. We planned to sleep at 12 and are currently up at 3AM because of deploy issues with APIs failing and the like.
Accomplishments that we're proud of
We’re proud of Halberd’s flexible design philosophy, which makes it easy for users to define custom behaviors and reuse moderation rules across servers.
The intuitive UI, smooth server integration, and insightful analytics dashboard are all results we’re particularly excited about.
What we learned
Throughout this project, we learned a tremendous amount about:
- Building fullstack applications with React and FastAPI
- Integrating and managing Discord bots
- Working with OpenAI models
- Implementing secure authentication with Google OAuth 2.0
- Designing scalable, user-friendly analytics systems
What's next for Halberd
Looking ahead, we have ambitious plans for Halberd:
- Tool-Augmented AI Detection: Integrate web search capabilities to help Halberd stay up-to-date with new slang and cultural trends.
- Audio Moderation: Extend moderation to voice channels by analyzing live audio conversations.
- Platform Expansion: Bring customizable moderation agents to more platforms like Reddit, GroupMe, and multiplayer games.
Halberd is just getting started-- we're excited to keep building the future of smart, customizable moderation!
Log in or sign up for Devpost to join the conversation.