environment-orchestratorlisted

# Environment Orchestrator An L2 Truth-Execution skill. Before any test runner can exercise scenarios, **the environment has to be real** — ephemeral databases, message queues, mock services, feature flags, seed data. This skill emits the recipe that makes the environment reproducible, pinnable, and tear-downable. It does not itself provision cloud resources; that's the `dev-ops` MCP's job. Thinking otherwise is how a "setup skill" ends up quietly leaking cloud accounts. ## When You're Invoked - **PIPELINE-3 step 1** — before `e2e-test-writer` / `uat-executor` run against a new fixture environment. - **On demand** as `/vibeflow:environment-orchestrator <profile> <platform>`. - **From `regression-test-runner`** when an integration suite needs a stood-up environment its local test harness can't provide. ## Input Contract | Input | Required | Notes | |-------|----------|-------| | Test type (profile) | yes | One of `unit / integration / e2e / uat / perf`. Loaded from `references/environment-profiles.md` §1–§5. | | Platform | yes | `web / ios / android / backend`. Some profiles don't apply to some platforms — the skill refuses `perf / ios` outright (iOS perf has its own toolchain, out of scope). | | Target environment name | yes | `--env <name>` — resolves to `vibeflow.config.json.environments[name]`. Production is rejected at the same layer as `uat-executor` (no override). | | Project config | derived | `vibeflow.config.json` for secrets path, environment URLs, doma