ZonicIsAlive
← all projects

chatbot

Discord bot utilizing LangGraph for advanced agentic reasoning, OpenRouter for LLM access, and ChromaDB for persistent long-term memory.

Python★ 0updated Aug 29, 2026view on github ↗

🤖 Autonomous Discord AI Agent

A state-of-the-art Discord bot utilizing LangGraph for advanced agentic reasoning, OpenRouter for LLM access, and ChromaDB for persistent long-term memory.

🚀 Advanced Features

  • 🧠 Agentic Reasoning (LangGraph): Unlike basic bots, this agent uses a state-graph to decide when to use tools, when to summarize conversation history, and how to maintain context across long sessions.
  • ⚡ Dual-Trigger System:
    • Slash Commands: Modern, discoverable UI commands.
    • Prefix Commands: Traditional ! commands for power users.
    • Conversational Triggers: Automatically replies to mentions, replies to its own messages, and DMs.
  • 💾 Evolving Memory:
    • Vector Store: Remembers your preferences and past conversations using ChromaDB.
    • Vibe System: Tracks a "Relationship Score" with every user that influences the bot's tone and helpfulness.
  • 🎵 Multimedia & Vision:
    • Vision Support: Can "see" and describe images you upload.
    • Music System: Full voice-channel integration with YouTube/SoundCloud support.
    • Image Generation: Built-in DALL-E/Flux support with automatic NSFW filtering based on channel settings.
  • 🛡️ Proactive Safety:
    • Nudge System: Detects heated arguments (5 messages in 10s) and attempts to de-escalate.
    • Age Verification: One-time mandatory verification for DM users to control NSFW content access.

🛠️ Command Reference

User Commands

SlashPrefixDescription
/help!helpInteractive help menu
/recap!recapAI summary of the last 100 messages
/ping!pingCheck bot latency
/status!statusCheck system health
/ask_memory-Query what the bot remembers about you
/vibe_with_me-Check your relationship score

Admin & Staff

  • /auto_reply <add/remove/list>: Manage channels where the bot replies automatically.
  • /personality <persona>: Override the bot's character for the current channel.
  • /vibe_check: Analyze the overall sentiment of recent server messages.
  • !sync: Force-synchronize all slash commands with Discord.
  • /set_model: Switch AI providers or models (e.g., GPT-4o to Gemini) at runtime.

🏗️ Architecture

The bot's "brain" is a LangGraph State Machine:

  1. Input: Receives message + Vision data + User Relationship score.
  2. Node: Agent: Decides if a tool (Music, Image Gen, Memory) is needed.
  3. Node: Tools: Executes requested actions and returns results to the agent.
  4. Node: Summarize: Every 15 messages, the agent automatically "compresses" its memory to stay fast and token-efficient.
  5. Output: Sends a natural, contextual response.

⚙️ Configuration

The bot uses a hybrid configuration system:

  • .env: Stores sensitive API keys and core model selections.
  • config.json: Stores all runtime settings (auto-reply channels, relationship scores, culture trends). This file is updated automatically by the bot.

Initial Setup

  1. pip install -r requirements.txt
  2. Configure .env with your DISCORD_TOKEN and OPENROUTER_API_KEY.
  3. Start the bot: python main.py.
  4. Use /auto_reply add in a channel to enable automatic chat.

🔐 Privacy

  • Conversations are stored locally in chroma_db/ for memory features.
  • User data is never sent to third parties except for the AI processing (OpenRouter/OpenAI).

📜 Credits

  • AI was somewhat used to assist in the creation and optimization of several features within this project.