Run a scenario

Pick a domain and search type. The engine runs on this server and saves results under /runs/<id>/ — you do not upload files. When the job finishes, links open the replay or Pareto viewer against those server paths.

Demo guide · How to Use (CLI) · Algorithms

What you get

Search type	Main output	Also saved
Find worst-case scenarios	Replay file → opens in the Replay viewer	Stripped-down replay (if the system collapsed), run logs
Coupled institution (research fork)	Replay + Pareto chart for the same GA budget	Run logs
Attacker vs. defender	Trade-off chart → opens in the Pareto viewer	Best-round replay, run logs

All files are saved under /runs/<id>/. A run index survives server restarts; individual run folders may be removed during maintenance.

Limits on this server

Maximum run time: 150 seconds — the run is stopped after that.
Maximum concurrent runs: 1. Extra submissions get a "busy" error.
Maximum 8 runs per hour per client address (rate limit).
Operators may require an API key for submissions (X-Fragility-Run-Key header).
All inputs are validated; no direct server access is possible.
For longer searches, larger populations, or local use: pip install fragility-engine — see How to Use.

Recent runs

Loading\u2026