TASK SUCCESS RATE
0%NO DATA YET
Waiting for first experiment
AGENT RELIABILITY / LAB 01
Make failure measurable.
Stress-test agent workflows, compare recovery strategies, and find the failure modes before production does.
ENGINE ONLINE v0.1.0
View source INVARIANT VIOLATIONS
0GUARDRAILS READY
Across displayed runs
RECOVERY ACTIONS
0AWAITING RUN
Retries observed in runs
TRACE EVENTS
0NO TRACE YET
Recorded execution events
EXPERIMENT STUDIO
Configure a run
Choose a fault scenario and put recovery policies to the test.
Fetching scenarios from the API…
RECOVERY POLICIES Select one or more
Repeat each policy for comparison.
3
Runs execute on the configured AgentFault API.
PERFORMANCE SNAPSHOT
Policy comparison
Success and invariant health by strategy.
Task success Invariant health
No recoveryNo runs
Blind retryNo runs
State-awareNo runs
Your next insight starts here
Run the same scenario under multiple policies to compare their behavior.
SCENARIO LIBRARY
Fault scenarios
Controlled failure modes shipped with AgentFault Lab.
Loading scenarios from your API…
EXECUTION TELEMETRY
Trace explorer
Inspect the event stream from the most recent individual run.
No trace selected
Run an experiment to inspect tool calls, fault injection, recovery and invariant events.