generated: '2026-09-19'
method: searched
source: https://relmcrm.com/docs (Base URL & auth; Webhooks) + https://relmcrm.com/security (Test and live are separate) + https://relmcrm.com/llms.txt (Key facts) + https://relmcrm.com/ (live test-mode widget)
checked: '2026-09-19'
summary: >-
  Relm's sandbox is a key-selected mode on the production host, not a separate environment: a relm_test_ key
  writes to an isolated test dataset on the same base URL, same endpoints, same MCP server. Test mode is free,
  unmetered, invisible to billing and to the live dashboard, and every record carries a mode field so data
  never crosses. It is explicitly a sandbox rather than storage — test records auto-delete 7 days after
  creation. OAuth tokens act on live data only, so the sandbox is reachable with API keys alone.
modes:
  selection: the key prefix decides the mode — relm_live_ (live) or relm_test_ (test); no header or separate host
  base_url_same: true
  base_url: https://api.relmcrm.com/v1
  mcp_same_endpoint: true
  key_prefixes:
    live: relm_live_
    test: relm_test_
  minted_at: https://app.relmcrm.com/
  keys_shown_once: true
  keys_at_rest: SHA-256 hash (docs + security page)
test_mode:
  free: true
  metered: false
  counts_against_quota: false
  visible_in_live_dashboard: false
  data_isolation: every object carries a mode; test and live datasets never mix (https://relmcrm.com/security)
  retention: test records (contacts, companies, deals, activities) are automatically deleted 7 days after creation
  oauth_available: false
  oauth_note: OAuth grants act on live data; test mode stays API-key only (https://relmcrm.com/docs, OAuth 2.1)
  webhook_targets: test-mode keys may point webhooks at localhost; live-mode URLs must be public https (private, loopback and link-local hosts are rejected)
try_without_account:
  available: true
  where: https://relmcrm.com/ — an in-page widget POSTs api.relmcrm.com/v1/contacts "for real, in free test mode" with no account or card
test_values:
  published: false
  note: No fixed test identifiers, magic values or fixture data are published; test mode uses ordinary requests against an isolated dataset. Nothing here is invented.
time_simulation:
  available: false
  note: No test clocks. Activities accept occurred_at to backdate imported history, which is a data field rather than a clock.
dry_run:
  available: partial
  operations:
  - operationId: previewSequence
    tool: relm_preview_sequence
    note: '"Dry-run: who would enroll now + the send schedule." No side effects. The only dry-run primitive on the surface; core object writes have no dry-run flag.'
fixtures_and_tooling:
  first_party: none beyond the SDK (npm relmcrm) and the batch endpoint
  resources: https://relmcrm.com/resources publishes pipeline/field/sequence/automation JSON templates described as valid against the API