| name | payment-risk |
| description | End-to-end provisioning of a payment-risk host on Alvera. Currently a STUB — the canonical setup walk lives in `tests/payment_risk/_order.json` at `@alvera-ai/platform-sdk@0.8.0`, which today contains only a smoke spec because the payment-risk industry is not yet productionised. Use this skill to get a stub setup conversation that lists payment-risk-specific resource types and routes to `/guided` for ad-hoc creation. Real domain orchestration replaces this stub when payment-risk integration coverage lands. Use when the user says "set up payment-risk", "provision a KYC/AML tenant", "fraud setup", or similar.
|
Payment-Risk host setup — stub
End-to-end domain orchestrator for payment-risk (KYC, AML, fraud).
Currently a stub. The integration-tests suite at
@alvera-ai/platform-sdk@0.8.0 includes only a smoke spec for
tests/payment_risk/; a real phase-by-phase walk will land here once
the payment-risk domain has full integration coverage.
SDK pin
This skill targets @alvera-ai/platform-sdk@0.8.0 exactly, matching
the rest of the domain skills. Payment-risk manifest:
https://github.com/alvera-ai/platform-sdk/blob/v0.8.0/integration-tests/tests/payment_risk/_order.json
Currently ["smoke"]. When real specs land, this skill will be expanded
to walk them in order.
What this skill does today
Routes the user to /guided for ad-hoc resource creation, with
payment-risk-specific suggestions. The primitive skills (guided,
custom-dataset-creation, DAC-upload, agentic-workflow-creation,
query-datasets) all work for payment-risk resources today — they're
domain-agnostic. What's missing is the opinionated phase sequence
that the healthcare skill provides.
Workflow
- Confirm SDK pin.
alvera --version should report 0.8.0. If
not, install @alvera-ai/platform-sdk@0.8.0.
- Set expectations explicitly. Tell the user this is a stub —
they're getting
/guided with payment-risk-flavoured suggestions,
not the full canonical walk that healthcare gets. If they need a
phase sequence today, point at the healthcare skill's
references/order-walk.md as a reference template they can adapt.
- Suggest typical payment-risk resource types. Common shapes:
- Custom datasets — KYC submissions, sanctions hits, AML
alerts, transaction-screening events. Use
/custom-dataset-creation.
- Tools — sanctions-screening APIs, identity-verification
vendors, internal alert webhooks. Use
/guided.
- AI agents + workflows — transaction triage, AML narrative
generation, sanctions hit review. Use
/agentic-workflow-creation.
- Hand off to
/guided for the resource loop. Don't try to
sequence — that's what the healthcare skill does and that machinery
isn't ready for payment-risk yet.
When this stub gets replaced
When tests/payment_risk/_order.json grows beyond ["smoke"],
re-pin this skill to the new SDK tag and replace this stub with a
phase-walk modelled on skills/healthcare/SKILL.md. The structure
should mirror healthcare's: SKILL.md + references/order-walk.md +
references/fixtures.md + references/example-transcript.md.
Out of scope
Same as the healthcare skill — prod tenant setup, cross-domain setup,
runtime ops are all out of scope here. Domain stubs aren't a license
to expand the surface.
