Tools

Extract recordings, transcripts, facts, documents, and templates from a corti-api tenant database.

Fact and document exports include both the generated and the current (clinician-edited) values:
facts carry original_text/original_source, document sections carry original_text, and guided
documents carry original_string_document/original_structured_document.

Template exports include template_versions (the version that produced each guided document),
template_version_sections (linking template versions to sections), and section_versions (the
published headings, instructions, and output schemas for each section). All walk template and
section inheritance chains via recursive CTEs, exporting all ancestor versions reached.

Downloads recordings from the voice-vault service using an Azure Bearer token.
Connects via kubectl port-forward and runs queries using a Docker postgres image.

Four modes:
  --tenant <name>                     Dump all data for the entire tenant
  --tenant <name> --user <id>         Extract all data for a specific user
  --tenant <name> --interaction <id>  Extract all data for a specific interaction
  --batch <csv>                       Batch extract from a CSV (headers: tenant,user_id,interaction_id)

--user, --interaction, and --batch are mutually exclusive. --batch does not require --tenant.

Prerequisites:
  - kubectl configured with the correct context and namespace
  - Docker running
  - Azure CLI logged in (az login) — required for recording downloads
  - op CLI installed and configured — required with --share-via-1password

1Password CLI setup (required for --share-via-1password):
  1. Install: https://developer.1password.com/docs/cli/get-started
  2. Open the 1Password app → Settings > Security → enable Touch ID, Windows Hello,
     or a Linux system authentication option
  3. Open the 1Password app → Developer > Settings → enable "Integrate with 1Password CLI"
  4. Sign in: op signin