Drop the pre-release dataset event resource and table now that Kobold edits are modeled as commit chunks.\n\nKeep record projections as the read model and expose compatibility events from commits at the API boundary.
Model dynamic dataset edits as synced commit chunks with bookmarks while keeping record projections as the local read model.\n\nPrivate/public visibility now only controls external tribe access; dataset metadata, resource definitions, and commits still publish through AshNostrSync for cluster-wide distribution.
Make Kobold require Trust, declare editable dataset/feature permissions, and enforce generic access rules for remote dataset reads and writes.\n\nUpdate the Docker e2e harness to include Trust and verify handshake-backed public access plus explicit private read/write grants.
Add a Nix-built two-node Docker e2e harness for Kobold that boots origin and edge Tribes nodes with the plugin installed.\n\nExpose admin management methods for the runner and assert public dataset sync while private datasets remain local.
Add fixed Ash resources for datasets, resource definitions, dataset events, and local record projections. Expose a loopback smoke API for public synced datasets and private local-only datasets, with migrations and host-backed tests.
self