Use case · Audit & evaluation

Benchmark the work, not the answer.

Whether an agent "works" can't be judged from its final answer, a wrong process can still produce a plausible-sounding conclusion. Knowing takes the full record: what it did, which evidence it used, which tools it called, what it ignored, and where a human intervened. Sentinel produces that record as a side effect of running.

evaluation · seeded scenario

Scenario · known root cause

Deploy introduces timeout on battery-state dependency

Ground truth
Found the deploy?Yes · ev-08Pass
Found the timeout trace?Yes · ev-09Pass
Checked stream health?SkippedGap
Diagnosis correct?Yes · citedPass

What the trail unlocks

Evaluation and audit from the same record

The full execution trace

Every tool call and output, every model prompt and response, every artifact and decision, in order, on an append-only ledger. The unit of evaluation is the whole investigation.

Ground-truth scenarios

Seed an incident with a known root cause and check whether the agent found the right evidence, not whether its answer merely sounds right. Grade process, not prose.

Reconstruction, months later

An auditor can replay any investigation as one causal story, trigger to recommendation, long after everyone involved has forgotten the details. No silent edits, ever.

Human interventions, included

Where a person confirmed, challenged, or took over is part of the record. That's how you measure trust honestly, including the cases where the agent needed help.

The method

From "seems smart" to measured

01 · SEED

Run scenarios with known facts

Dispatch a realistic incident, alert-shaped or incident-shaped, against seeded observability data where the root cause is known in advance. The agent investigates it exactly as it would a live incident.

scenario.ts
// dispatch a ground-truth incident
await sentinel.events.dispatch({
  sample: "ingest-5xx-after-deploy",
  truth:  { cause: "timeout", evidence: ["deploy", "trace"] }
})
→ investigation runs as if live
02 · GRADE

Score the trace against the truth

Did it find the deploy? Did it pull the right trace? What did it skip? Because the trail records everything, including omissions, the gaps are as visible as the hits, and they become the next iteration's fix.

grade · run 14
Required evidence found · 2 of 2 pass
Stream health check · skipped gap
Diagnosis matches ground truth pass

Evaluate agents the way you'd audit them.

Ground-truth scenarios, full execution traces, and a record that answers the hard questions long after the incident is closed.