Rensei docs
Factory

Factory Dashboard

Software Factory Floor dashboard overview.

The Software Factory Floor dashboard is a real-time analytics view of your SDLC pipeline, showing throughput, cycle time, cost, and code quality metrics across all stations (research, development, security, QA, acceptance, deploy).

Overview

The Factory dashboard displays a comprehensive summary of your software development process, aggregating work items, costs, and completion metrics across your configured SDLC workflow. It is the primary analytics view for workflow editors and operators who need to understand factory performance.

Time Windows

All metrics support three time ranges:

  • 7 days - week-over-week comparison, immediate trends
  • 30 days - monthly performance baseline (default)
  • 90 days - quarterly planning and long-term trends

Clicking the time-range selector in the top toolbar updates all panels simultaneously.

Screenshot placeholder: A screenshot of the full Factory dashboard with all five panels visible would help readers orient before they read each panel's detailed page.

Dashboard Composition

graph LR
    A[Factory Dashboard\n7d / 30d / 90d] --> B[Executive Summary\nTotal Issues · Cost · Cycle Time · Yield]
    A --> C[Station Pipeline\nresearch→dev→security→qa→accept→deploy]
    A --> D[Cost Breakdown\ncost/station · trend · human vs agent]
    A --> E[Code Survival\n7d/30d/90d survival + routing reward]
    A --> F[Performance\nvia Performance tab]
    A --> G[Memory Analytics\nvia Memory sub-page]

Panels Overview

The dashboard is composed of five main panels:

1. Executive Summary

A KPI card row showing:

  • Total Issues - count of completed work items in the time window
  • Total Cost - sum of all costs (in USD, formatted to 2-3 decimal places)
  • Avg Cycle Time - median time from issue open to deploy (human-readable: s/m/h/d)
  • Avg Yield - percentage of first-pass merges (0-100%)

These four metrics are your most important health indicators. A healthy factory typically aims for high yield (>90%) and low cycle time (<24h for most stages).

2. Station Pipeline

A left-to-right flow visualization of the six SDLC stations:

  • research → development → security → qa → acceptance → deploy

For each station, the panel displays:

  • WIP count - number of items currently in that station (if > 0)
  • Throughput - items completed per day
  • Cycle Time - median time spent in that station
  • Yield - percentage of items passing (first time, no rework)

Stations are color-coded by yield health: green (≥90%), yellow (70-89%), red (<70%).

3. Cost Breakdown

Two-panel view showing cost distribution:

Left panel:

  • Avg Cost per Issue - the primary cost metric
  • Cost per Station - bar chart showing which stages are most expensive
  • Cost Trend - sparkline of daily cost over the time window

Right panel:

  • Time Distribution - split between human-authored and agent-authored work
  • Throughput - items completed per day (efficiency metric)

Cost is denominated in USD and reflects the sum of LLM provider costs, compute (sandbox), and estimated human time. It is calculated from factory_metrics table with type cost_per_issue.

Using the Dashboard

  • Rising cost, flat throughput → improving quality or longer deliberation per item
  • Flat cost, rising yield → better first-pass rates; agents are learning
  • Rising throughput, rising cycle time → saturation; increase capacity or parallelize
  • Declining yield, stable WIP → quality regression; check provider, SDLC template, or agent skill

Drilling Deeper

  • Click on a station to drill into station-specific analytics (planned)
  • Hover over cost bars to see per-station breakdown
  • Use the project dropdown to filter to a specific project (if multi-project org)

Time Windows

The time-range selector affects all metrics globally:

  • 7d shows day-by-day granularity; useful for daily standup
  • 30d (default) shows weekly patterns and is the standard planning window
  • 90d shows quarterly trends and seasonal variance

Data Freshness

Metrics are updated every hour via the factory aggregation job (factory/aggregation-job.ts). Real-time event ingestion (work items, handoffs, rework) is reflected in the events stream immediately, but aggregated metrics (cost, yield, cycle time) have a 1-hour lag.

API Access

Programmatic access to all dashboard data is available via the Metrics API:

curl -H "Authorization: Bearer $RENSEI_API_KEY" \
  "https://app.rensei.ai/api/factory/metrics?metricType=aggregate&timeRange=30d"

See the Metrics API reference for full details.

Executive Summary

Totals, cost, cycle time, and yield.

On this page

OverviewTime WindowsDashboard CompositionPanels Overview1. Executive Summary2. Station Pipeline3. Cost BreakdownUsing the DashboardInterpreting TrendsDrilling DeeperTime WindowsData FreshnessAPI AccessRelated Pages