Platform
ScaiWave ScaiGrid ScaiCore ScaiBot ScaiDrive ScaiKey Models Tools & Services
Solutions
Organisations Developers Internet Service Providers Managed Service Providers AI-in-a-Box
Resources
Support Documentation Blog Downloads
Company
About Research Careers Investment Opportunities Contact
Log in

Changelog

User-visible changes only. Internal refactors and infrastructure work omitted.

v2 — NTFS-style ACLs#

The big change. Per-collection and per-document ACLs with allow/deny ACEs, transitive group expansion, owner concept, and single-chokepoint search filtering.

  • Eight-verb permission bitmask (READ, WRITE, DELETE, INGEST, LIST, READ_PERMISSIONS, CHANGE_PERMISSIONS, TAKE_OWNERSHIP) plus composed roles (VIEWER, EDITOR, MANAGER, OWNER).
  • Allow and deny ACEs with NTFS resolution order; deny wins at the same level.
  • Parent inheritance with per-resource break (PATCH /permissions/.../acl with inherit_from_parent: false).
  • Ownership transfer (POST /permissions/.../take-ownership), audit-logged.
  • Transitive group expansion via a ScaiKey nested-group mirror table and a ten-minute reconcile cron, with webhook fast-path.
  • filter_results_by_acl chokepoint — every search, list, and graph result passes through it. Property-tested across 30 random scenarios.
  • Admin UI PermissionsPanel mounted on collections and documents.
  • v1 collection-level /access endpoints retained for backward compatibility.

v1 — Launch#

First generally-available release.

  • Collections with per-collection embedding model, chunking strategy, and counters.
  • Five chunking strategies: fixed, paragraph, semantic, markdown, code.
  • Document ingestion by upload, URL, ScaiDrive, and bulk multipart.
  • Website crawl with depth, page-count, and byte budgets; recurring crawl configs with daily / weekly / cron schedules; HMAC-verified inbound webhook trigger; SSE progress stream.
  • Vector, hybrid, and keyword search; global and per-collection variants.
  • Graph extraction (optional per collection); manual node / edge CRUD; query, traverse, path, ask, context, search; live SSE stream of mutations; NDJSON export and bulk import; cluster summary and changes-since for large-graph UIs.
  • Re-chunk and re-extract jobs for in-place updates; fork for embedding-model changes.
  • Saved graph views (private and tenant scope).
  • Admin pages: Knowledge Dashboard, Collection Manager, Document Browser, Search Playground, Graph Explorer, Ingestion Monitor, Crawl Manager.
Updated 2026-05-18 15:01:30 View source (.md) rev 12