Overview Panel

The Overview panel (O) is the main analysis dashboard. It opens on the left side of the workspace and holds four tabs that share one canvas: Overview, Stability, Monte Carlo, and Sweep.

When to use it

Watch the system run a single sim and see the live aggregate metrics (Overview).
Investigate frame-to-frame consistency separately from delivery success (Stability).
Run N replications with varied seeds and look at the distribution (Monte Carlo).
Sweep one or two behavior knobs across a grid and find the right config (Sweep).

The four tabs

Overview

A snapshot of the current run:

Health — composite 0-100 score derived from delivery, drop, and queue pressure. Color codes: green (healthy), amber (warning), red (critical).
System Profile — radar chart of throughput, latency, retention, recovery, headroom. Quick visual for “what shape is this design”.
Queueing Analysis — Little’s Law variables (lambda, L, W, Lq, Wq, rho) plus the max-rho hotspot pointer that names the worst-utilized component. Configurable warm-up window (skip first N transient frames so steady-state metrics aren’t polluted).
Critical Bottleneck — names the worst component with peak queue and lost-packet count, plus a recommended fix (capacity bump or parallel split).
Performance — utilization / delivery / active-step / queue-peak progress bars per component.
Side-by-side baseline diff — save a baseline, change the design, see percentage deltas color-coded by lower-is-better.

Stability (Pro)

Frame-to-frame consistency, NOT delivery success. A system can deliver 100% of requests (Healthy in Overview) and still behave erratically here, or stay perfectly stable while everything drops.

Verdict — predictable / erratic / unstable.
Instability score — 0-100, lower is better.
Health weights — adjust how much throughput / drops / queue weight into the verdict.
Insights — auto-detected patterns (e.g. “queue oscillates between empty and full”).
Top variance components — which blocks contribute most to the instability.
Timeline sub-tab — step-by-step health table.

See the dedicated Stability Analysis page for the verdict thresholds, the health formula, and how the smoothing for small runs works.

Monte Carlo (Pro)

Runs the same sim N times with varied seeds and aggregates the distribution.

Run controls — pick run count (5-100). Each run gets a different seed.
Distribution metrics — avg / best / worst health, percentiles, std dev.
Per-component reliability — peak queue P5/P95 ranges, drop rate, processing volume per component.
SLO compliance — set min delivery rate / max dropped / min health and see what fraction of runs pass.
Seed list — preserved in the repro bundle export so anyone can rerun and get the same numbers.

See the dedicated Monte Carlo page for run-count guidance, SLO setup, and how seed variation affects results.

Sweep (Pro)

Parameter sweep with a 2D heatmap. Pick a behavior knob (capacity / serviceRate / split-weight), set min / max / steps, run Monte Carlo per cell, get a color-coded grid showing how throughput / dropped / peak utilization / mean health change as the knob moves.

See the dedicated Parameter Sweep page for setup details and worked examples.

Copy for Docs

One button in the panel header (next to the close X) bundles Overview + Stability + Monte Carlo into a single plain-text snapshot. Paste straight into a PR description, Slack thread, design doc, or paper writeup. The snapshot is plain text (not Markdown) so it lands cleanly in any tool.

The button is hidden on the Sweep tab (sweep results aren’t included yet — copy them from the export modal’s Citation bundle instead).

Keyboard shortcuts

O — toggle the Overview panel
Tab buttons in the panel header switch between the four sub-tabs
+ / - on the warm-up frames stepper adjust the queueing aggregate window