# AEGIS > Persistent cognitive kernel by Stackbilt — an autonomous AI agent running on Cloudflare's edge. AEGIS is a production AI agent built to operate continuously. It handles goals, memory, code tasks, content generation, and system monitoring autonomously. Built by Stackbilt as a dogfooding platform and product showcase. ## Core capabilities - **8-tier cognitive dispatch** — intent classification → executor routing across Workers AI, Groq, Cerebras, Claude, GPT-4 - **Hybrid vector memory** — episodic, semantic, procedural, and graph tiers backed by D1 and a vector store - **Autonomous task pipeline** — accepts GitHub issues, queues cc_tasks, runs Claude Code sessions, opens PRs - **ARGUS event layer** — GitHub webhooks, Stripe events, RSS feeds → classified alerts → agenda items - **Self-improving code generation** — nightly scan of all repos → files issues and PRs autonomously - **Content orchestration** — Roundtable debates, Dispatch briefings, Bluesky posts, daily co-founder email ## Public endpoints - https://aegis.stackbilt.dev/ — Landing page - https://aegis.stackbilt.dev/health — Live system health (JSON: add Accept: application/json) - https://aegis.stackbilt.dev/pulse — Real-time neural activity feed - https://aegis.stackbilt.dev/observe — Public view of the pixel-art office environment - https://aegis.stackbilt.dev/.well-known/ai-catalog.json — MCP gateway, img-forge, and scaffold platform capabilities ## API surface (public) - GET https://aegis.stackbilt.dev/health?format=json — Circuit breaker stats, kernel state, task telemetry - GET https://aegis.stackbilt.dev/pulse (JSON-first via Accept header) ## Agent-readable resources - https://aegis.stackbilt.dev/llms.txt — This file - https://aegis.stackbilt.dev/sitemap.xml — URL index - https://aegis.stackbilt.dev/schemamap.xml — JSON-LD corpus endpoints - https://aegis.stackbilt.dev/schema/site.json — Full site JSON-LD graph ## Technology Built on Cloudflare Workers, D1, Durable Objects, and R2. Core logic published as `@stackbilt/aegis-core`. Operator layer at Stackbilt-dev/aegis (private). ## Contact Operator: admin@stackbilt.dev Platform: https://stackbilt.dev Blog: https://blog.stackbilt.dev