RLTasksby ulam.ai
Sign inAll tasks
Operations & simulationsIncident-Commander

Incident-Commander

A compositional incident-response environment for diagnosing and mitigating outages in a synthetic multi-region microservice system. Episodes combine initiating mechanisms, dependency edges, regional or tenant scope, propagation, observability failures, recovery blockers, convergence stages, and plausible unrelated changes; agents must gather evidence, apply narrow controls, verify recovery independently, communicate accurately, and submit an evidence-linked postmortem.

Decision makingStateful environmentRLVR
Version3.0.0
Environments1
RewardScalar · 0–1
DeliveryPrivate ZIP
01 Task contract

A real, versioned RL environment.

A compositional incident-response environment for diagnosing and mitigating outages in a synthetic multi-region microservice system. Episodes combine initiating mechanisms, dependency edges, regional or tenant scope, propagation, observability failures, recovery blockers, convergence stages, and plausible unrelated changes; agents must gather evidence, apply narrow controls, verify recovery independently, communicate accurately, and submit an evidence-linked postmortem.

Agent objective

  • Interact with the supplied stateful environment.
  • Produce verifier-checkable actions or artifacts.
  • Maximise scalar reward under the package contract.

Evaluation

  • Private verifier.
  • Reported reward range 0–1.
  • Package-specific public and private checks.

Delivery boundary

  • Private object stored in Cloudflare R2.
  • Authenticated entitlement required.
  • Short-lived signed URL per download.
02 What you will work on

Distinct environments, one demanding research contract.

Enough detail to understand the intellectual terrain; generated instances, hidden mechanisms, and solution paths remain inside the private package.

EnvironmentMathematical or technical frontierAdaptive research problem
Release and configuration failuresSemantic regressions, treatment skew, schema mismatchSeparate causal changes from plausible decoys, infer scope, and choose a narrow rollback, isolation, or compatibility plan.
Capacity and dependency failuresTenant pressure, pools, retries, cachesTrace propagation across caller–callee edges, discover hidden residual state, and avoid broad damaging controls.
Data-plane poisonQueues, offsets, payloads, write contentionQuarantine exact bad state, clear compatible blockers in order, and verify that lag and secondary effects converge.
Identity and discovery failuresCertificates, secrets, DNS, clock skewReconcile incomplete telemetry and misleading changes before applying precisely scoped rotation or routing controls.
Regional convergenceFailover, shard, lease, compressionRecover every affected region, validate synthetic journeys and traces, observe stabilization, communicate accurately, and close an evidence-linked postmortem.
03 Why it is interesting

What the supplied evaluation reveals.

A blind expert trajectory restored service, found the exact root cause and residual blocker, and scored 0.9447—but failed because its first identified update preceded blocker discovery. That is compelling evidence that recovery alone cannot compensate for premature communication. Across 64 calibration cases the reference passed all and five negative policies passed none; 512 generated incidents were structurally unique. A disclosed rejected-action clock-mutation bug must be fixed before replay-perfect production claims.

We publish aggregate behavior and task structure, while withholding generated instances, hidden labels, exact successful probes, private checks, and solution trajectories.

04 Supplied evaluation

Observed evaluation result.

Shown with its provenance and limitations; it is not a performance guarantee.

i
Methodology matters

Reported result from the evaluation artifact supplied with this package.

Evaluated system / policyGPT-5.6 Pro-assisted evaluation

As identified by the supplied artifact.

Mean reward0.9926

512 reported runs.

Result artifactIncluded

incident_commander_v3_test_results.zip:incident_commander_v3_test_summary.json

Public result record

Machine-readable provenance and the exact displayed metric are available in results.json.

Open result JSON
05 Private delivery

The package stays off the public website.

The paid ZIP will live in a private R2 bucket. Vercel authorizes the buyer and issues a 2–5 minute object URL; R2 serves the bytes directly.

Included with purchase

  • Exact package version 3.0.0
  • Environment and task contracts
  • Verifier or scoring interface
  • Supplied reference/evaluation artifacts
  • Purchase record and licence v1.1
delivery flow
Authenticated buyer + entitlement check
+ private R2 object + 2–5 minute signed URL
= direct, auditable download

Package SHA-256
1c05330138254bba1806e68895bc799443bedb39175e19faec280cf2b6881844
06 Licence v1.1

Commercial use, without exclusivity.

One purchase licenses this identified item to one legal organisation for worldwide, perpetual commercial model training, evaluation, research and development. Redistribution and resale of the package are not permitted.

Read the full licenceYotta Content LTD · business customers only
Incident-Commander

Ready to add this environment?

Back to marketplace
Stripe checkout

Business purchase confirmation

Sign in or create an account, then complete secure Stripe Checkout. Access is granted only by the verified payment webhook.