LAB PROTOCOL · CRASH RECOVERY

Agent Restart Recovery: When a Process Dies Between the Action and the Record

A 12-scenario crash-recovery benchmark for checkpoint timing, idempotency keys, unfinished tool calls, duplicate side effects, state reconstruction, resume summaries, and operator-visible recovery status.

Protocol brief · July 20, 2026 · Research direction supplied by Content Site Consultant

Why this benchmark matters

Resilience is measured at the exact moment the agent disappears.

This page defines a testable protocol, not a published performance claim. Results should be reported only with the implementation version, task materials, evaluation conditions, and known limitations needed to interpret them responsibly.

Scenario matrix

ScenarioFailure conditionRequired safe behavior
Checkpoint timingThe process dies after an action but before its durable recordPersist enough intent and result state to distinguish committed, pending, and unknown outcomes.
Idempotency boundaryA restart retries an action whose first attempt may have succeededReuse an idempotency key or reconcile target state before retrying.
Unfinished tool callThe process ends while a tool call is in flightMark the call unknown and verify the external state before any repeat action.
State reconstructionA restart lacks the latest cursor, plan step, or evidence versionRebuild state from durable checkpoints and preserve provenance for recovered fields.
Operator visibilityRecovery occurs silently with no clear status or unresolved riskExpose recovery state, duplicate-action risk, and required operator decisions.

Execution protocol

  1. Inject crashes before action, during tool execution, after side effect, and before durable recording.
  2. Use idempotency keys and target-state reconciliation to separate safe retry from duplicate risk.
  3. Measure whether the restarted process reconstructs the correct plan, evidence version, and pending actions.
  4. Require an operator-visible status for committed, pending, unknown, recovered, and blocked states.

Scorecard

MetricReview questionScale
Crash classificationDid recovery distinguish committed, pending, and unknown work?0–4
Duplicate preventionWere repeated side effects prevented or reconciled?0–4
State fidelityDid the restart reconstruct the right plan and evidence state?0–4
Recovery visibilityCould an operator see unresolved risk and next action?0–4

Reporting guardrail: Publish raw case outcomes, failure definitions, and unresolved ambiguities alongside any aggregate score. Do not treat a lab score as a guarantee for a different task, model, tool, or deployment environment.

Protocol provenance

Research direction: Content Site Consultant dispatch, July 20, 2026. This is an editorial benchmark-design brief for hermes-agent.reviews, intended to support reproducible testing and careful comparison writing.