Tour: Dashboard
Your orientation — the loop-first home you land on. What’s pending in the review loop comes first; the health signals sit beneath as supporting context. Two modes: Monitor (the default) and Onboard (an on-demand on-ramp for a repo you don’t know yet).Monitor mode
The default view once you’re past initial setup.- Review loop tile — surfaced first when your agent has flagged something: open agent flags, notes awaiting your reply, and last agent activity, with a click straight to Review. When nothing’s pending but you’ve used the loop, it turns into a calm “all caught up” tile — the cockpit isn’t only ever alarms
- Dispatch ledger — the cockpit’s instrument panel: every fix that’s out, returned, or pending — for this repo and across every other repo you have open in a tab. Each row clicks straight to the file in Review. Hidden entirely when nothing’s in flight. When work is out, a Copy prompt button composes one kickoff prompt covering both queues — your dispatched goals and your fix-requested review notes — and copies it for you to paste into your own agent (or tell an MCP-connected agent to “work Reikon’s dispatch queue” and it pulls the work itself). The agent fixes in its own working tree and the diff returns here to Review; Reikon never runs or applies anything itself
- Health signals — the six-signal breakdown (bus factor, commit quality, TODO density, duplicates, churn stability, dependency health), each scored 0–100, plus an overall label (Healthy / Needs attention / At risk / Critical) and a one-line “why” explaining the single biggest factor behind the score. See Formulas & Methodology for the exact math behind every number here.
- Language bar — proportional breakdown of lines by language
- Trend sparklines — appear once you have 2+ saved snapshots: commit quality, TODOs, bus factor, health score, and test coverage (if configured) over time
- Stale branch alert — collapsible card listing branches with no activity in 30+ days
- Contributor risk alert — bus-factor files (single owner ≥60%) whose owner hasn’t committed in 30+ days. The bus-factor concern made concrete: not just “ownership is concentrated” but “concentrated, and the person has gone quiet”
- Most changed files / top contributors — quick-glance cards, click through to the full view
- Compare — opens a Then→Now delta view between any two snapshots, with an optional AI-generated narrative summarizing what changed
Onboard mode
Built for someone new to the codebase — a new hire, or you revisiting a project after months away. Three sections instead of a full signal breakdown:- Start here — low-complexity files with known improvement opportunities (existing TODOs or tech debt) — safe places to make a first contribution
- Talk to these people — top 3 contributors by recent (30-day) activity, with the directories they work in most
- Avoid for now — high-churn, high-complexity files — not literally forbidden, just risky to touch without context
- Understand first — files with single-person ownership, high complexity, and no recent commits. The owner who could explain it might not remember the details either by now — read before you touch, not after