generated: '2026-07-17'
method: searched
source: >-
  https://github.com/AltScore/altscore-python (src/altscore/__init__.py) and
  https://github.com/AltScore/altscore-cli (.claude/skills/altscore-api/SKILL.md)
description: >-
  AltScore's test surface. There are two complementary mechanisms: (1) a fully
  separate sandbox environment (distinct auth/API/borrower-central hosts) that
  the SDK and CLI select via an environment name; and (2) an in-production test
  mode (isTest) that lets UAT records coexist with live data. No public test
  card / magic-value list is published. Never invent test values.
environments:
- name: production
  hosts:
    identity: https://auth.altscore.ai
    borrower_central: https://bc.altscore.ai
    altdata: https://data.altscore.ai
    cms: https://api.altscore.ai
- name: sandbox
  hosts:
    identity: https://auth.sandbox.altscore.ai
    borrower_central: https://bc.sandbox.altscore.ai
    cms: https://api.sandbox.altscore.ai
- name: staging
  hosts:
    identity: https://altscore-stg.us.frontegg.com
    cms: https://api.stg.altscore.ai
- name: local
  note: URLs supplied via ALTSCORE_LOCAL_* env vars.
selection:
  sdk: AltScore(environment="sandbox")
  cli: profile/environment chosen during `altscore login`
test_mode:
  field: isTest
  detail: >-
    Most entities support isTest so UAT data can live in production. Test records
    are excluded from list results by default.
  cli_flags:
    include_tests: --include-tests
    only_tests: --test-only
    create_as_test: --is-test
    mark: altscore <resource> set-test <id> --enable   # cascades to children
    unmark: altscore <resource> set-test <id> --disable
  full_test_mode_entities:
  - borrowers
  - identities
  - documents
  - deals
  - assets
  - borrower-fields
  - deal-fields
  - asset-fields
  - points-of-contact
  - deal-contacts
  - authorizations
  - metrics
  - artifacts
  - data-models
  - evaluators
  - evaluation-rules
  - policy-rules
  - rule-trees
  filter_only_entities:
  - executions
  - execution-batches
notes: >-
  DRAFT v2 workflows execute their full graph (node execution is not gated by
  status), so running a DRAFT by id is a real pre-publish test; publishing only
  changes which version the alias serves.