OmniMod

OmniMod: a moderation dashboard that gives Reddit mods a centralized command center for tracking, reviewing, and reverting moderation action
Real-time Action Ledger with category tabs, status filters, and stats. Track every removal, ban, warning, and spam action.
Select actions and bulk revert with AI safety scanning. One click to undo moderation mistakes across multiple posts.
AI Provider settings with Gemini & OpenAI support. Generate warning templates and analyze rule gaps with one click.
Configurable ledger size, AI scan sensitivity levels, and Markdown archive export for long-term mod accountability.
AI-generated warning templates from subreddit rules. Consistent mod messaging with manual template support.

Inspiration

Moderation mistakes are inevitable in active communities, especially when a team is handling removals, spam, bans, warnings, appeals, and policy edge cases at high speed. What inspired OmniMod was the gap between Reddit’s powerful moderation logs and the day-to-day workflow moderators actually need: a single place to review what happened, understand patterns, and safely undo mistakes.

What it does

OmniMod is an AI-powered moderation command center for Reddit. It gives moderators one dashboard for tracking moderation actions, reviewing removals and warnings, auditing team activity, and bulk-reverting mistakes from a custom Devvit post inside the subreddit.

Core features include:

Live Action Ledger: Captures removals, spam flags, bans, and warnings from Devvit onModAction events.

Bulk Revert Engine: Lets mods select multiple logged actions and approve/revert them in one fault-tolerant workflow.

OneVoice AI Warning Templates: Generates consistent, rule-aware warning text with optional Google Gemini or OpenAI BYOK configuration.

Rule Gap Analysis: Finds recurring moderation patterns that may indicate missing or unclear subreddit rules.

Safe-to-Revert AI Scan: Optionally checks selected removed content for threats, doxxing, CSAM indicators, or malware before restoration.

Mod-Only Dashboard: Enforces moderator-only access server-side and keeps data scoped to the subreddit installation.

How we built it

OmniMod was built with Devvit Web, TypeScript, Hono, Redis, Reddit events, and a custom HTML/CSS/JavaScript dashboard. Devvit event handlers capture moderation actions, Redis stores subreddit-scoped records, and the webview provides a mod-only interface inside a custom Reddit post.

For AI features, OmniMod supports optional BYOK integrations with Google Gemini and OpenAI, so communities can generate warning templates and review safety signals without requiring a hardcoded platform key.

Challenges we ran into

The hardest part was making bulk reverts safe and reliable. A moderation action can fail for many reasons, so the app needed to handle each revert independently instead of stopping the whole batch.

Another challenge was balancing AI usefulness with moderator control. OmniMod uses AI as an assistant for warnings, rule gaps, and safety checks, but the moderator remains in charge of final action.

Accomplishments that we're proud of

I’m proud that OmniMod feels like a real moderation workstation rather than a demo. It has a live action ledger, filters, counts, bulk operations, AI tools, settings, and export workflows. It also works directly inside Reddit, which keeps moderators in their normal context instead of sending them to another service.

What we learned

I learned how to build a complete Devvit Web app with event-driven moderation data, Redis-backed state, custom webviews, moderator-only access, and optional AI integrations. I also learned that moderation tooling needs to optimize for trust: every automated helper should be explainable, reversible, and scoped to the community.

What's next for OmniMod

Next, I would add richer appeal workflows, team notes, per-rule analytics, scheduled moderation reports, and more granular permission controls for larger mod teams. I would also expand the AI safety layer so moderators can tune it to each community’s rules and risk tolerance.