AI SRE · Intermediate · FICTIONAL ENVIRONMENT
Agent Loop Containment
An incident-analysis agent repeatedly calls two diagnostics tools and never reaches a terminal state.
MISSIONStop runaway autonomous execution and recover task completion.
MISSION PROGRESS
0 / 4
PHASE 1
Open evidence sources. Build your incident picture.Investigate
INVESTIGATION JOURNAL
Outputs appear here as you inspect systems.Observed evidence
Evidence you inspect will appear here.
PHASE 2
This is not scored until you choose to check it.Form a root-cause hypothesis
PHASE 3
Controls unlock after an evidence-consistent hypothesis.Apply a mitigation
PHASE 4
Do not declare victory before verification.Verify recovery
Debrief
- Autonomous workflows need deterministic budgets.
- Tool semantics should support progress and terminal states.
- Observe actions and outcomes rather than private chain-of-thought.