Choose the lab that matches the proof you need.

The three lab types are intentionally different. A control drill tests a decision. A brief specifies work to build. The runnable lab lets you execute and inspect working code.

Interactive · 3 minutes · beginner

Control decision drill

Inject one failure and identify the first trusted control to exercise. No setup and no code.

Prerequisite: noneKeep: one fail-closed responseType: interactive decision exercise
Try the drill →

Written · 45–90 minutes each · intermediate

Five technical lab briefs

Build and prove a bounded implementation slice. These are written briefs, not starter repositories or runnable labs.

Prerequisite: related missionKeep: implementation evidenceType: written build brief
Read the briefs →

Runnable · 60–90 minutes · intermediate

Claims-denial proof lab

Run a zero-dependency Node 22 service and exercise candidate binding, authorization, duplicate safety, readback, and recovery.

Prerequisite: Node 22 + HTTP basicsKeep: test and recovery traceType: runnable proof lab
Open runnable lab →

Choose the smallest safe route.

Change the failure. The drill shows the first control to exercise; the technical briefs show how to build deeper proof.

Failure injection

What failed?

Choose the observed symptom before reaching for a framework or agent topology.

Fail closed: reject the unsupported result, preserve the source and trace, and route the case for review.