Mission Control

Mission Control is the home view of the Agents workspace in the macOS app. It's a single screen that answers "what's happening in my vault right now?" without requiring you to dig through individual panels.

What it shows

Mission Control is composed of summary cards followed by panels. The exact items vary based on what's actually present in your vault, but the structure is consistent.

Summary cards

A row of high-level counters at the top of the view:

CardShows
Connected agentsHow many MCP clients have an active session
Pending proposalsNumber of AI-suggested changes awaiting your review
Recent activityTool calls in the last 24 hours
Patterns detectedCount of derived patterns surfaced in the Patterns view
Vault healthA 0–100 score derived from orphan count, pending proposals, and recent errors
Orphan pagesPages with no inbound or outbound wikilinks

The vault health score is computed as 100 − (orphan_pages × 2) − pending_proposals − (recent_errors × 3), clamped to a minimum of 0. The breakdown is visible in the Vault Health view if you want to see exactly what's pulling it down.

Recent activity

A compact list of the most recent MCP tool calls — what tool, by which session, with status and duration. Click through for the full activity log in Sessions & Activity.

Active sessions

A summary of MCP clients that are currently connected — client name, protocol version, auth method (OAuth or API key), and time of last activity. Click through to Sessions & Activity for the full list.

Pending proposals

The proposals waiting on you, with their type and a one-line summary. Click through for the full proposal review in Proposals.

Vault health snippets

A condensed view of the most important integrity findings — recent orphans, recent broken-link reports, and any recurring error from agent activity. The full picture lives in Vault Health & Permissions.

How to use it day to day

Mission Control is built for quick checks, not deep work:

  • Once a day — glance at it to see what your AI clients have done.
  • Before approving a proposal — Mission Control flags new proposals before you have to dig.
  • When something feels off — the vault health score and recent errors will surface anything broken before you have to investigate manually.

Everything Mission Control shows is also accessible through MCP tool calls (mindwiki_get_agent_activity, mindwiki_list_agent_sessions, mindwiki_list_proposals, mindwiki_lint) or REST endpoints — so the same data is available to your AI when you ask it to summarize what's happened.

Where to go next