AEGIS Documentation
Agent Execution, Graph Intelligence & State
AEGIS is a stateful runtime platform for enterprise AI compliance agents. Compliance domains are installable packs, entity types are runtime-defined, and agent behavior — skills, routers, rules, prompts, and settings — is tenant-configurable data rather than code. Every regulatory action runs through mandatory human-in-the-loop governance and lands in an append-only audit trail.
The Texas Railroad Commission (RRC) oil & gas vertical is the first installed pack set — five packs (Rule 37 spacing, Rule 32 flaring, Form PR production reporting, Flaring Monitor, and EPA OOOOb methane) covering spacing exceptions, flaring compliance, portfolio monitoring, and emissions tracking. The mechanisms are domain-neutral: a new vertical is a new pack, not a fork.
Choose Your Documentation
Developer Docs
Architecture deep-dives, local setup, service reference, database schema, testing, and contribution guidelines for engineers building AEGIS.
User Guide
How to use the AEGIS platform — compliance dashboards, filing workflows, agent conversations, entity management, and flaring monitoring.
API Reference
Authentication, rate limiting, endpoint reference, SSE event schemas, and per-service OpenAPI specifications for integrators.
Quick Facts
| Architecture | 11 services (10 Python + 1 Go), Next.js 16 frontend |
| Agents | Seeded general + creator; domain behavior arrives via skill routers and packs |
| Compliance Domains | Installable packs (the RRC oil & gas vertical is the first installed) |
| Execution Engine | LangGraph stateful graphs |
| Knowledge Store | PostgreSQL + Apache AGE (graph) + pgvector (embeddings) |
| Caching / Memory | Redis 7 |
| Messaging | Apache Kafka |
| Governance | HITL approval checkpoints, append-only audit trail, RBAC |