Context Engineering for DevOps: Turn Your Infra into a Graph
summary
Claims that a live knowledge graph reconciles cloud state, Terraform code and remote state to detect drift and shadow IT, built by any shift and implemented with a graph database (examples: Neptune) and an explicit ontology. Ingest uses cloud APIs and CloudTrail, mapping AWS resources to Terraform entities, producing graphs that scale to millions of nodes with versioning. Also demonstrates LLM limits with GitHub copilot and short-term fixes via MCP servers while noting MCP API size limits for large infra.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:01The Importance of Context in Production Systems
- 1:14Challenges with Infrastructure as Code
- 3:14Limitations of AI in Infrastructure Code
- 5:54Introduction to Graph Databases
- 8:11Building and Populating the Knowledge Graph
- 9:48Using MCP Servers for Context
- 11:10Creating a Mapping Between Cloud Data and Terraform Code
- 13:09Conclusion: The Key Role of Context
related talks
29mGraphs: The Fabric of DevOps
07mTalk Deep Dive
44mEvolution of Infrastructure Management
40mHow Terraform’s simple graph concepts deploy complex infrastructure as code
24mTerraform Best Practices
48mHashiConf Day 2 Keynote- 40mHow Terraform’s simple graph concepts deploy complex infrastructure as code
32mCloud Computing 101