Valora

Valora is an eve-based, Groq-powered Discord moderation and multipurpose community bot.

What Is Included

• Groq model configuration through @ai-sdk/groq.
• Native eve Discord channel at POST /eve/v1/discord.
• Always-on Autonomous Community OS instructions.
• On-demand skills for moderation policy and community operations.
• Sixteen agentic feature tools for moderation, security, support, community operations, analytics, missions, and Shipyard.
• Typed tools for content classification, case logging, trusted autonomy-gated Discord actions, and poll drafting.
• Specialist subagents for moderation, raids, scams, policy, support, community ops, vibe analysis, and Shipyard.
• Vitest unit tests plus Eve eval smoke coverage.

Setup

1. Copy .env.example to .env.local and fill in the values.

2. In Discord Developer Portal, set your Interactions Endpoint URL to:

   https://your-domain.example/eve/v1/discord

3. Register slash commands. During development, guild commands propagate faster.

curl -X PUT "https://discord.com/api/v10/applications/$DISCORD_APPLICATION_ID/commands" \
  -H "Authorization: Bot $DISCORD_BOT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "name": "ask",
      "description": "Ask Valora for community help",
      "type": 1,
      "options": [
        {
          "name": "message",
          "description": "What should Valora do?",
          "type": 3,
          "required": true
        }
      ]
    },
    {
      "name": "mod",
      "description": "Ask Valora to review a moderation issue",
      "type": 1,
      "options": [
        {
          "name": "message",
          "description": "Message, report, or incident details to review",
          "type": 3,
          "required": true
        }
      ]
    }
  ]'

Autonomous Community OS Commands

Valora exposes these Eve tools for Discord slash-command prompts and agentic use:

• /sentinel - autonomous moderation review.
• /quarantine - temporary restriction workflow.
• /risk - user trust and risk profile.
• /case timeline - incident timeline.
• /appeal - appeal intake and recommendation.
• /raid shield - raid protection workflow.
• /scam scan - scam and malware analysis.
• /policy learn - structured policy learning.
• /explain action - member-safe moderation explanation.
• /modbrief - moderator briefing.
• /vibe check - community health analysis.
• /answer - FAQ and support answers.
• /announce - announcement drafting.
• /mission - community quests.
• /ops - broad multi-step operations.
• /shipyard - playful friend-server chemistry analysis with opt-out.

Tier 2 actions can run autonomously only when enabled by guild config and when confidence meets the configured threshold. Kicks, bans, permanent role changes, critical-severity cases, and final appeal decisions remain approval-gated.

The real Discord action tool also requires VALORA_AUTONOMY_ACTION_TOKEN for any approval bypass. Keep this token server-side; do not paste it into prompts, Discord messages, or public logs.

Free-Tier Hosting Shape

The MVP is designed for:

• Vercel Hobby for Eve HTTP interactions.
• Neon Free for future durable Postgres memory through DATABASE_URL.
• Upstash Free for future Redis cooldowns and burst windows.
• Groq free/rate-limited usage through GROQ_API_KEY.

The current local fallback stores Valora state in data/valora-state.json. Production deployments should use durable hosted storage before relying on autonomous moderation at meaningful scale.

Local Development

npm exec -- eve dev

The Eve CLI is interactive, so this repo does not start it automatically.

Verification

npm run typecheck
npm test
npm run build
npm exec -- eve eval --list

Full eval runs call the live model and require GROQ_API_KEY.