OpsMill Agentic Operations for Infrahub
Networking Field Day 41
•
20m
Alex Gittings of OpsMill demonstrated agentic operations with Infrahub, using AI to query a network’s intent data and request service changes through governed workflows. The session covered an MCP server that learns each customer’s schema, permission pass-through and branch-based guardrails for agent actions, reusable skills that enforce process, and AI-assisted schema development using OpsMill’s skills and spec-driven tooling. Gittings presented the final segment of OpsMill’s appearance at Networking Field Day 41.
Gittings explained that the MCP server runs as a sidecar to Infrahub and learns each customer’s schema, including descriptions of why relationships exist. That context lets an AI assistant reason over the intent data as a true knowledge graph. In the demo, he asked whether any security policies covered the lab’s Grafana instance. The assistant reported that port 3000 was open only to the pod network, not the branch. He then asked it to grant branch access using a skill, which he said speeds up lookups, saves tokens, and enforces a consistent runbook-style process. The MCP server acts with the requesting user’s own credentials and SSO-assigned role, so an agent can do no more than its user. Write access is off by default, and even when enabled, the server always creates a branch and never pushes to main. It checks that configurations rendered correctly before handing off to a human approver. Delegates pressed on telling a person apart from that person’s agent. Gittings acknowledged the current design passes the user through, and delegates suggested hardware keys as one way to prove a human was involved.
In the second part of the demo, Gittings showed how OpsMill’s skills help builders extend Infrahub. Using Claude Code, he asked for a per-device monitoring flag, and the skill added a documented Boolean attribute with a default value to the device schema. The skills work with any AI tool and install via npx. He said they handle implementation but don’t replace an operator’s design judgment. For larger services, OpsMill’s spec kit extension supports a spec, plan, approve, and build workflow. Infrahub also flags conflicts when multiple branches change the same data.
Presented by Alex Gittings, Solution Architect. Recorded live at Networking Field Day 41 in Santa Clara, CA on October 8, 2026. Watch the entire presentation at https://techfieldday.com/appearance/opsmill-presents-at-networking-field-day-41/ or visit https://techfieldday.com/event/nfd41/ or https://OpsMill.com/ for more information.
Up Next in Networking Field Day 41
-
HPE Networking Data Center Evolving A...
AI infrastructure is rapidly evolving from centralized training environments into complex inference-driven systems in which a single workload can traverse front-end, back-end, storage, and security fabrics. HPE Networking’s strategy is to deliver open, secure Ethernet for every AI fabric (scale-u...
-
HPE Networking Data Center Scale-up N...
AMD Helios delivers massive, rack-scale AI architectures, backed by HPE expertise in liquid cooling, system integration, and high-performance networking. Helios effectively announces the arrival of Ethernet into the scale-up domain, which has been dominated by proprietary protocols. Customers wil...
-
HPE Networking Data Center Exploring ...
Managing congestion in AI training and inference fabrics is critical to maximize GPU utilization, minimize latency and jitter, and protect model accuracy and throughput. Ultimately, congestion-free networks are required to deliver the best end user experiences in the era of agentic AI. This sessi...