The use case is a single loop you can describe in order—no parallel reality where the “plan” lives in one place and the product ships another.
Why this beats a spreadsheet-only process
A manual plan in a spreadsheet splinters fast: different tabs, one-off exports, and email threads that each claim to be the source of truth. The sheet usually sits next to the repo, not in the same permission and change model as your product—so the app ships one story and the “official” list lags or forks by default, unless someone re-enters every change by hand.
A governed catalog holds one set of definitions: who can change what, what is live or retired, and a trace you can trust when questions come later. MCP means your IDE is not maintaining a shadow copy in a file no one else sees—it works on the same contract as the app, so you get speed without off-book edits that never rejoin the team.
You start with a workspace and tokens so people and tools only see what they should. You connect MCP in your own environment (for example from your IDE) so the catalog is not a copy you edited offline; it is the same definitions, the same rules, the same permissions as in the app.
Then you shape the plan where you already work. You move names, properties, and structure until the catalog is what you are willing to stand behind—before the next release trains more drift into dashboards and experiments. When it is ready to build against, you pass it on: teammates in the workspace, read-only access where that is enough, and the EventPanel CLI so what developers generate matches what was approved, not a loose interpretation of an old message thread.
Governance sits in the middle of that loop, not at the end of a footnote. Review, audit, and lifecycle (what is live, what is retired) keep both humans and automation inside the contract—so speed does not cost you trust in the data later.
EventPanel is where your tracking plan becomes a governed catalog — and your IDE talks to it by MCP.