Skip to Content
v0.8.0 · shippedNative iOS / Android / Flutter / Capacitor SDKs, A2A discovery, SOC 2 readiness, residency, BYO storage, BYOK. Read the changelog →
Admin consoleAction inbox
One primary CTA per stage — what to do next · · open live demo ↗

Action inbox

Route: /inbox

Scenario: It’s Monday morning. You open Mushi and want to know in one glance what actually needs a human decision right now — without clicking through Reports, Queue, Health, and Drift separately to piece it together.

The Inbox aggregates every outstanding action from across the PDCA pipeline into a single prioritised list. Think of it as your daily standup prep sheet, auto-generated.


How it works

The Inbox reads entirely from GET /v1/admin/dashboard — no extra API calls, no extra latency. Every card you see is a state in the pipeline that Mushi has identified as requiring a human decision or awareness.


Reading an action card

Each card has four parts:

PartExample
PDCA stage chipPLAN (triage backlog), DO (failed fix), OPS (dead-letter item)
Page label”Reports”, “Queue”, “Integration health”
Action title”5 reports awaiting triage”
Reason”Oldest is 18 hours old — over your 12-hour SLA”

The primary action button takes you directly to the page with the relevant item pre-selected. You never need to search.


Cleared stages strip

Below the open cards, a row of chips shows every pipeline stage that currently has nothing pending — with a ✓ badge. This is deliberate: it’s a positive signal. When all chips are checked and there are no open cards, your pipeline is clear.

Click any cleared chip to navigate to that page and verify the state yourself.


Filtering by PDCA phase

Use the filter chips to focus on one part of the loop:

FilterShows
AllEverything open
PlanTriage backlog — reports needing a human decision
DoFailed fix runs, stuck PRs
CheckQuality failures — judge drift, health probe errors
ActPromotable lessons, unattributed releases
OpsDead-letter items, cron failures

Common tasks

Morning triage (the fast path)

  1. Open Inbox. Look at the Open count in the top-left tile.
  2. If it’s 0 — all cleared. You’re done. Dashboard is the next stop.
  3. If not, scan the cards top-to-bottom — they’re ordered by severity.
  4. Click the primary action on each card to resolve it.
  5. Refresh (button top-right) — cleared cards disappear as you resolve them.

Understanding a stage chip colour

  • Red chip — critical severity or SLA breach — do this first.
  • Amber chip — warning — action today.
  • Grey chip — informational — awareness only.

When the Inbox seems wrong

If you resolved an issue but the card is still showing: click Refresh. The Inbox reads from the dashboard cache (60-second TTL). The card will clear on the next poll.


The Inbox is a read-only view — it doesn’t take actions itself. Every primary action button navigates to the page that owns the action. This keeps the audit trail clear: actions are logged on their source page, not here.


Last updated on