Before a table moves to a new model, you size the rewrite and say where to start. By hand, that's every column, one at a time: what writes it and what reads it, across plugins, flows, form scripts, column mappings, AI prompt columns and an agent. Then which of them feed each other, so you know what to touch first.
Pathix starts from the table. The logic footprint has one row per component: what it writes, what it reads, and how Pathix knows. In the demo environment that's plugin steps, cloud flows, a form script, column mappings, Dataverse functions, AI prompt columns, a Copilot Studio agent and its knowledge source. Fifteen are Deterministic, and the agent is Unresolved.
The rewrite plan puts the components in five waves. Wave zero is safe to change first: nothing downstream depends on what it produces. The plugin from Ep. 03 is the last wave: it creates every engagement, and the components around it read what it writes. Tracks that share nothing can move at the same time.
Three components sit in wave zero with no column edge: nothing is known to depend on them, which isn't proof that nothing does. Two links from the agent are marked unproven, because Pathix couldn't resolve one of its writes, so they don't get to pin the order. The plan is a labeled direction, not an asserted build order. Verify it before you build to it.
One Pathix call returns the waves, with every flag still on them. With the Dataverse MCP alone, your agent rebuilds the footprint first, one component at a time, and the plugins come back compiled.
A read-only metadata scan of your environment that reads the logic inside the components: compiled plugins, flow definitions, workflow XAML and scripts.
How it works →Or start by hand: what Show Dependencies leaves out→All questions →
A 30-minute walkthrough on a pre-scanned demo environment. No access to your tenant, nothing to install.