Skip to content

Vespera — The Silent Architect

A modular, AI-powered Discord bot for Cloud Engineering, Tabletop RPGs, and Community Moderation — designed to run on 1 vCPU / 1GB RAM.

  • <2s Latency

    Near-instant responses via Groq LPU inference for time-critical game interactions.

  • 1-Core / 1GB VPS

    Aggressive memory optimizations — GC tuning, String Interning, strict LRU caches — keep the footprint flat over time.

  • 186 Cloud Best Practices

    RAG knowledge base across GCP, AWS, and Azure powering the Cloud Engine advisor.

  • RAG-Powered Accuracy

    The Truth Block Protocol constrains AI output to verified facts only — no hallucinated spells or invalid Terraform resources.

What is Vespera?

Vespera is a Discord bot built around a "Persona as Infrastructure" philosophy. Rather than bolting AI responses onto a simple command router, every module — Cloud, D&D, Moderation — shares a single VesperaPersonality singleton that guarantees consistent tone, color palette, and error messaging across all 15+ cogs.

The system dynamically routes tasks to the best model for the job: Groq (Llama-3) for sub-500ms game round responses and Gemini Pro 1.5 for large-context cloud document analysis. All AI outputs touching factual domains (D&D rules, Terraform specs) are constrained by Truth Blocks — verbatim data injected from verified sources that the model is strictly instructed to format, never invent.

View Architecture → GitHub Repository

Tech Stack

Layer Technology
Runtime Python 3.11, asyncio
Bot Framework discord.py
Database SQLite (WAL mode)
Fast Inference Groq — Llama 3.3 / Mixtral
Large Context Google Gemini Pro 1.5
IaC Generation Terraform (AWS / GCP / Azure)
Data Store SQLite (bot_database.db, cloud_infrastructure.db, cloud_knowledge.db)

Core Modules

Module Purpose
☁️ Cloud Engine AI infrastructure advisor, Terraform code generation, cost estimation
🐉 D&D System 5e rules engine, character management, narrative AI
🛡️ Utility Core Translation, summarization, automated moderation