Can AI agents automate Documaster Arkiv?

Documaster Arkiv is automatable today through its documented Noark 5 HTTP/JSON API. The v1 API exposes transactional writes, a query language, document upload, and document download across the archive-core model: fonds, classification, series, folders, case files, registry entries, documents, national identifiers, and shared metadata. Writes go through POST /noark5/v1/transaction as an ordered save, link, unlink, or delete sequence that rolls back on any error. Reads use POST /noark5/v1/query. Authentication is OAuth 2.0 client credentials against the instance IDP at /oauth2/token, with one-hour access tokens and an openid scope. Credentials, certificates, and hosts are issued during onboarding; the API is server-to-server only. Official Java and .NET client libraries are published on Maven Central, NuGet, and GitHub. Access groups assigned to the integration user control read, write, and update rights, and Noark plus GDPR rules still govern every write. Documaster lists more than 70 standard line-of-business integrations, including KS FIKS Arkiv and Microsoft 365 add-ins; those are product integrations, not Zapier, Make, Power Automate, or n8n connectors. Those four catalogs had no Documaster app on 2026-08-21. Vendor documentation does not settle first-party MCP ownership, and the official MCP registry had no Documaster or Noark server. No general CLI or computer-use probe was established. Contract management and Microsoft 365 add-in products are outside this archive-core boundary except where they use this API.

Best path today: api · Overall automatability: 7/10