Getting Started with Project Aegis

Welcome to Project Aegis - an autonomous AI agent platform designed for continuous operation with minimal human intervention.

What is Project Aegis?

Aegis is an AI agent that: - Operates 24/7 within a contained Linux environment - Routes tasks to appropriate LLM tiers (Claude, GLM-4.7, local models) - Maintains persistent memory across sessions (episodic, semantic, knowledge graph) - Communicates via Discord, WhatsApp, Telegram, and email - Runs scheduled jobs for monitoring, research, and automation - Hosts live products (Intel Dashboard, Research API, Open Notebook)

System Overview

Environment

• Host: Hetzner EX130-R dedicated server
• Container: LXC with 110GB memory limit
• Services: Docker Compose stack (5 containers)
• Database: PostgreSQL with 137 tables + FalkorDB knowledge graph

Live Services

Service	URL	Purpose
Dashboard	aegisagent.ai	Main interface
Intel	intel.aegisagent.ai	Geopolitical intelligence
Notebooks	notebooks.aegisagent.ai	Research tool
Code	code.aegisagent.ai	VS Code editor

Communication Channels

Channel	Purpose	Access
Discord	Primary status, logs, tasks	Guild 1454722052777836546
WhatsApp	Mobile commands	+447441443388
Telegram	Time-sensitive alerts	Chat 1275129801
Email	Triage and automation	aegis@richardbankole.com

Quick Start

1. Check System Health

# Via WhatsApp
status

# Via Discord
/status

2. Access the Dashboard

Visit aegisagent.ai for: - System status overview - Recent activity feed - Product catalog - API documentation

3. Run a Command

# Morning routine
/morning

# Research a topic
/research best practices for API rate limiting

# Check emails
/email-triage

Key Concepts

Cognitive Hierarchy

Aegis routes tasks to the most cost-effective model:

Tier	Model	When Used
1	Claude Opus 4.5	Strategic decisions (rare)
1.5	Claude Haiku 4.5	Fast ops (classify, extract)
2	GLM-4.7	90%+ of work (~8 req/min)
3	Ollama Local	Fallback, offline, vision

Memory Architecture

Episodic   → Events, decisions, interactions
Semantic   → Knowledge, facts, learnings
Procedural → Workflows, how-tos
Journal    → Daily entries (YYYY-MM-DD.md)
Graph      → FalkorDB entity relationships

Work Priorities

1. SHIP & ACTIVATE (70%): Wire features, fix bugs, validate
2. REACTIVE WORK (20%): Discord tasks, system health
3. PROACTIVE WORK (10%): Research, documentation

Daily Schedule

00:00-06:00 UTC  Maintenance (backups, updates)
06:00-08:00 UTC  Morning check (status, planning)
08:00-22:00 UTC  Active work
22:00-24:00 UTC  Evening summary

Next Steps

1. Quick Start Guide - 5-minute system orientation
2. First Commands - Essential commands to know
3. Architecture Overview - How it all works
4. Operations Guide - Day-to-day operations

Getting Help

• Discord #tasks: Post task requests
• Discord #alerts: Monitor system alerts
• WhatsApp: Send help for command list
• Email: aegis@richardbankole.com

Philosophy

Aegis operates on these principles:

1. Enable behaviors, don't ship features - Build what changes user behavior
2. Chase latent demand - Look for unarticulated needs
3. Ship first, perfect later - 70% shipped beats 100% unshipped
4. Three-strike protocol - Retry → local reasoning → escalate
5. Resource discipline - $50/month budget, tier-aware routing

---

Ready to dive deeper? Continue to Quick Start.