---
license: cc-by-4.0
language:
- en
task_categories:
- question-answering
tags:
- sre
- kubernetes
- root-cause-analysis
- agents
- it-operations
pretty_name: ITBench-AA
size_categories:
- n<1K
configs:
- config_name: sre
  data_files:
  - split: test
    path: sre/data.jsonl
---

# ITBench-AA

Artificial Analysis' release of the public scenarios from
[IBM's ITBench benchmark](https://github.com/itbench-hub/ITBench), used for
the [ITBench-AA leaderboard](https://artificialanalysis.ai/evaluations/itbench-aa-sre).

This repo currently contains the **SRE** subset (`sre` config). Each row is a
Kubernetes incident scenario with its expected contributing-factor entities. An
agent under evaluation is given access to an offline snapshot of the affected
cluster (alerts, events, traces, topology) and must identify the entity
(Deployment, Pod, ConfigMap, etc.) responsible for the failure.

The `sre` config contains only the 40 scenarios marked
`source_split: public` in our internal dataset (out of 59 total). The
remaining private/held-out scenarios are not included.

## Source

The scenarios originate from [`itbench-hub/ITBench`](https://github.com/itbench-hub/ITBench).
See the per-scenario directories at
`scenarios/sre/project/roles/scenarios/files/scenario_<N>` in that repo.

## Fields

- `id_aa`            — Artificial Analysis row identifier
- `scenario_id`      — upstream scenario identifier (e.g. `Scenario-3`)
- `category`         — module identifier (`sre`)
- `source_split`     — always `public` in this release
- `scenario_root`    — filesystem root used inside the evaluation sandbox
- `ground_truth_yaml` — YAML describing the fault, alerts, entity groups,
  propagation graph, and recommended remediations for the scenario

## Citation

If you use this dataset, please cite IBM's original ITBench paper alongside this
release.