Reference
Architecture, configuration, enterprise features, and migration guides for the AI Templates stack.
Technical reference for the three-repo AI Templates stack after the Deep Agent merge (August 2026).
Reference guides
| Guide | Description |
|---|---|
| Stack architecture | Full-stack diagram, ports, and request flow |
| Config-as-code | config/agent/ vs config/ui/settings.yaml |
| Enterprise features | Langfuse, OTEL, SSO, MCP auth, OPA, HITL — by repo |
| Migration from pre-Deep Agent | Breaking changes checklist for adopters |
Template documentation
For hands-on setup, see the template sections:
Workshop
- Zero to Production: Agentic AI Onboarding — end-to-end local stack walkthrough
Announcements
External resources
AI Templates builds on open specifications and frameworks. These upstream docs explain concepts used throughout this site:
| Resource | What it covers |
|---|---|
| LangGraph Deep Agents | Deep Agent orchestration model used by template-agent |
| LangGraph documentation | Graph runtime, streaming, checkpoints, LangGraph API |
| Agent Skills specification | SKILL.md format for skills in config/agent/skills/ |
| Model Context Protocol | MCP tools and server integration (template-mcp-server) |
| Langfuse | Optional tracing and feedback (template-agent) |
Our template repositories remain the source of truth for how these are wired in this stack—ports, config paths, and deployment overlays are documented in the guides above.
For questions, visit GitHub Discussions.