Skip to content

Status Dashboard

The dashboard provides a glanceable status view for monitoring AI agent runs across all your projects. Access it at / when the web UI is running.

Time to orient: Under 30 seconds to understand what needs attention.

Accessing the Dashboard

Start the web UI with rp1 arcade:

rp1 arcade
# Opens http://localhost:7710

Data Sources

The dashboard displays real data from your local rp1 installation. All run data is sourced from the status database at ~/.rp1/rp1.db, which is populated when AI agents report their progress using the emit agent tool.

How Runs Are Populated

When an agent (or any workflow) reports status via rp1 agent-tools emit, a record is written to rp1.db. The API queries this database to populate the dashboard:

  1. Agent executes workflow - Agents call rp1 agent-tools emit at key milestones (feature started, task in progress, completed, failed)
  2. Status stored in database - Each update creates a record in ~/.rp1/rp1.db with run ID, step, status, and event data
  3. API queries database - The /api/v2/runs endpoint queries the database for the latest status per feature
  4. Dashboard displays runs - The UI renders runs grouped by their current status

Status Value Mappings

The status database stores granular status values that are mapped to dashboard display statuses:

Database Status Dashboard Status Description
started Running Initial execution state
in_progress Running Active work in progress
waiting-input Waiting Agent blocked, needs user input
needs-review Needs Review Work complete, awaiting review
completed Completed Success terminal state
failed Failed Error terminal state

The dashboard groups runs into attention sections based on these mapped statuses:

  • Waiting for you - Runs with waiting-input status
  • Needs review - Runs with needs-review status
  • Failed - Runs with failed status
  • Running - Runs with started or in_progress status

Empty Dashboard

If your dashboard shows no runs, ensure that:

  1. Agents are reporting progress via rp1 agent-tools emit with validated state machine transitions
  2. The status database exists at ~/.rp1/rp1.db
  3. The project is registered (runs are filtered by registered projects)

Home Dashboard (Now View)

The home dashboard (/) shows what needs your attention across all projects. Runs are grouped into four sections displayed in priority order:

Section Description Default State
Waiting for you Runs blocked waiting for user input Always expanded
Needs review Completed runs with artifacts to review Always expanded
Failed Runs that encountered errors Always expanded
Running Active runs in progress Collapsed

Each run item displays:

  • Status badge with color and icon
  • Project and feature name
  • Command that was invoked
  • Current step (for running items)
  • Relative timestamp ("2 min ago")

Click any item to view full run details.

Empty State

When nothing needs attention, you'll see a calm message: "Nothing needs your attention. All your agent runs are proceeding smoothly."

Refresh

Click the Refresh button in the header to manually fetch updated data. The dashboard also receives real-time updates via WebSocket when runs change status.

Runs List

The runs list (/runs) shows all agent runs with filtering capabilities.

Filters

Filter Options
Status All, Running, Completed, Failed, Waiting
Project Dropdown of registered projects
Date Range Today, This Week, This Month, All Time

Filters combine with AND logic. Clear all filters with the "Clear filters" button.

Filter state syncs to URL parameters for shareable links.

Project-Scoped View

Navigate to /projects/:projectId/runs to view runs filtered to a specific project.

Run Detail

The run detail view (/runs/:runId) shows complete information about a single run.

  • Large status badge
  • Command invoked (e.g., /build, /pr-review)
  • Feature/branch name
  • Started time and duration
  • Breadcrumb navigation: Project > Feature > Run

Step Timeline

Horizontal timeline showing workflow progression:

Step State Visual
Completed Green checkmark
Current Blue filled circle (animated for running)
Pending Gray empty circle
Failed Red X

Agent Sub-State Panel

When a workflow step has associated agent activity, the step timeline displays a nested agent activity panel below the workflow diagram. This provides hierarchical visibility into what agents are doing within each phase.

What it shows: - Agent name (humanized, e.g., "Task Builder" for task-builder) - Current agent step and status badge - Per-task tracking: individual task IDs (e.g., T1, T2) with independent state per task

Behavior: - Steps with active (running) agents are auto-expanded - Steps without agent activity render identically to before (no visual change) - Updates in real-time via the existing WebSocket mechanism - Each task within an agent progresses through the agent's state machine independently

Example: During a build workflow's "build" phase, you might see: - task-builder: building (T2) -- with T1 shown as completed

See State Machines for how agents define and report state machine transitions.

Artifacts Panel

Lists files produced or updated during the run:

  • File type icons (markdown, diff, code, etc.)
  • File path
  • "New" badge for newly created artifacts

Click an artifact to open the file path.

Event Stream

Collapsible panel showing the chronological event log:

Event Type Description
step-start Step began execution
step-complete Step finished
warning Non-fatal warning
error Error occurred
artifact-updated File was created/modified
task-batch Summary of completed tasks

Task-batch events show a summary like "12 tasks completed in Build step" rather than individual task events.

The left sidebar provides navigation between views:

Item Route Description
Home / Attention dashboard
Runs /runs All runs list
Projects /projects Project list

The sidebar can collapse to icon-only mode for more content space.

Header

The header displays:

  • Logo and branding
  • Project switcher dropdown
  • WebSocket connection status indicator
  • Theme toggle (dark/light)
  • Help button (links to documentation)

Keyboard Shortcuts

The dashboard supports a comprehensive keyboard-first interaction model. See Keyboard Shortcuts for the full reference.

Global

Shortcut Action
Cmd/Ctrl + K Open command palette
? Toggle shortcut help overlay
/ Focus search input
Escape Dismiss overlay or blur focus
Cmd/Ctrl + \ Toggle sidebar collapse

Go-To Chords

Chord Destination
g then h Home
g then r Runs
g then p Projects

List Navigation

Key Action
j / Arrow Down Select next item
k / Arrow Up Select previous item
l / Arrow Right Drill into item
h / Arrow Left Drill out to parent
Enter Open selected item
Escape Clear selection

Keyboard navigation uses the roving tabindex pattern for accessibility.

Run Detail

Key Action
Escape Return to runs list

Theme

The dashboard uses the Catppuccin color palette:

Theme Description
Mocha (dark) Default theme with deep purple-blue tones
Latte (light) Light theme with warm cream tones

Toggle between themes using the sun/moon button in the header. Your preference is saved to localStorage.

Status Colors

Status Color
Queued Gray (Overlay)
Running Blue
Waiting Peach
Completed Green
Failed Red
Needs Review Mauve

Real-time Updates

The dashboard maintains a WebSocket connection for real-time updates. When an agent run changes status, the UI updates automatically without page refresh.

If the WebSocket disconnects, the connection status indicator shows "reconnecting" and a 5-second polling fallback activates. Updates resume normally when the connection is restored.