README.md

metadata

license: cc-by-4.0
task_categories:
  - text-generation
language:
  - en
tags:
  - devops
  - sre
  - software-engineering
  - benchmark
  - agents
  - mcp
  - deterministic-evaluation
pretty_name: DevOpsBench-100
size_categories:
  - n<1K

DevOpsBench-100

DevOpsBench-100 is a synthetic long-horizon software-engineering / SRE agent benchmark: 100 tasks over one executable world ("NovaCart", a mid-size e-commerce SaaS) with 72 SQLite tables, 1451 seeded rows, a 38-file monorepo with 417 commits, and 97 MCP tools spanning a first-party engineering stack (tickets, PRs, CI, deployments, canaries, migrations, feature flags, metrics, alerts, incidents, chat, knowledge base) plus deliberately disagreeing vendor-shaped surfaces (Jira, Linear, GitHub Issues, Prometheus, Sentry, PagerDuty, Confluence, spreadsheets) and Kubernetes.

Tasks are outcome-only tickets (symptom + definition of done; company policy lives in the world's knowledge base, not the prompt). Reference trajectories run 4-35 tool calls (median 13), with 100/100 distinct tool-name sequences. Every task publishes 14 task-scoped initial-state views, three operational alternatives, and 59-86 explicit vcode/evidence/containment criteria. Acceptance is fully deterministic: each task ships an executable vcode verifier that checks the final world state and the append-only audit log, with anti-forgery table pins - no LLM judge, no network, no clock in the reward path.

What is included

• data/tasks.jsonl: task records (task_id, task_name, world_id, prompt, context_files, rubric, gold_output, metadata).
• tasks/: one readable JSON record per task (includes the guided instruction variant).
• task_files/: 14 inspectable views per task spanning tickets, services, observability, incidents, deployments, code, CI, migrations, vendor trackers, knowledge, chat, approvals, and exact tool contracts.
• world/: the offline world source - stdlib MCP server, tool implementations, seeded SQLite database, schema and seed SQL.
• verifiers/: 100 standalone verifier scripts (python3 verify_<task>.py world.db prints the full verdict).
• trajectories/: one executed reference trajectory per task (JSONL).
• reports/: measured build and qualification evidence.

Task families (19)

Family	Tasks
Cross-system source of truth	8
Error-rate SLO recovery	8
Latency optimization	8
API migration	7
Change attribution	7
Feature-flag operation	7
Multi-service rollout	7
Security incident response	7
AIOps root-cause analysis	6
Flaky-test remediation	6
AIOps detection	5
AIOps localization	5
Cross-source reconciliation	5
Code implementation	4
Operational judgement / restraint	4
Incident handover	2
Long-horizon delivery	2
Human-approval gated change	1
Workspace scripting	1

Objective release gates

Gate	Required	Measured
Tasks	100	100
High-level unique employee requests	100	100
Unique reference tool sequences	100	100
Inspectable assets per task	>=12	14
Public criteria per task	>=40	59-86
Oracle replays at reward 1.0	100/100	see reports/qualification.json
Deterministic verifier replays	100/100	see reports/qualification.json
Negative-control false accepts	0	see reports/qualification.json
LLM / network calls in verifier	0	0

Provenance and contamination

Every service, metric, document, commit, and incident is synthetic and was generated for the NovaCart world. 30 cross_system/handover tasks were ported from TheAgentCompany task shapes re-grounded onto NovaCart's own state; the AIOps families reproduce the microsoft/AIOpsLab task structure (detect / localize / analyze) against NovaCart. No third-party benchmark text, evidence, or gold answers are included. Gold outputs are public, so this release is appropriate for transparent evaluation and RL experiments rather than secret-test claims.

Licenses

Synthetic task data and world content are CC-BY-4.0. Benchmark code, server, and verifiers are Apache-2.0.