Skip to content

MigrationOps Copilot

An AI safety net for cloud migrations: diff live source against target, diagnose the break, gate the fix behind a human.

My role
Builder · Microsoft AI Dev Days 2026
When
2026
Status
Microsoft AI Dev Days Hackathon
Stack
Python, FastAPI, Azure OpenAI

01

Where it started

Website migrations routinely ship hidden outages. DNS cutovers propagate unevenly, SSL certificates get forgotten on new load balancers and unmapped paths return silent 404s, and most teams still catch this with ad-hoc curl scripts.

02

What I owned

Builder · Microsoft AI Dev Days 2026

03

How I built it

  • Captured live SSL trust, DNS resolution and HTTP status and latency from source and target endpoints, then diffed them into a structured JSON report.
  • Chained bounded LLM stages (triager, diagnostician, planner) on Microsoft Agent Framework and Azure OpenAI, each reasoning only over captured data.
  • Gated a non-destructive, simulated executor behind explicit human approval, exposed through a CLI, a FastAPI web UI and an MCP server, with GitHub Actions CI.
[Source endpoint]        [Target endpoint]
        │                        │
        ▼                        ▼
 [Live snapshot: SSL · DNS · HTTP status/latency]
                    │
                    ▼
       [Deterministic JSON diff engine]
                    │
                    ▼
 [Triager] ─▶ [Diagnostician] ─▶ [Planner]   (bounded LLM stages)
                    │
                    ▼
     [Human approval gate: CLI / Web UI]
                    │
                    ▼
   [Simulated executor + verification trail]
System architecture
  • One-way pipeline (a DAG, not an open loop): network telemetry, then deterministic parsing, then bounded reasoning, then simulated mutation.
  • Risk is classified Critical/High/Medium/Low with a 0–100 health score, so engineers can triage at a glance.
  • Health checks are also published as an MCP server, so other agents can call them as tools.

Facts first, reasoning second

LLM stages only ever see the captured snapshot diff, never the open internet, which keeps diagnoses grounded in what was actually measured.

Execution is simulated by design

Approved plans go to simulated infrastructure handlers, so the tool can never make a bad migration worse.

04

What changed

A broken Azure App Service cutover is flagged critical, traced to a missing SSL binding and paired with a reviewable fix plan before users ever hit it.

PythonFastAPIAzure OpenAIMicrosoft Agent FrameworkMCPGitHub Actions