# Northstar Community Workshop - Type: Vercy Dimension - Preset: vercy.preset.organizational.community - Namespace: example.northstar-workshop - Owner: public-example:vercy - Created: 2026-09-05T20:49:51Z - Canonical location: R:\02_PROJECTS\02_Meta_Models_Platforms\Ver.cy\current\vercy-platform-experience\site\examples\reference-workshop\dimension - Vercy entry: https://ver.cy/AGENTS.md - Presets: https://ver.cy/presets/index.yaml - Model catalogue: https://ver.cy/models/ - Runtime model index: https://ver.cy/models/runtime-index.json - Model resolution API: https://ver.cy/api/v1/models/resolve/ - Model request registry: `registries/model-requests.yaml` - Model protocol: https://ver.cy/model-agent-protocol.md - Local query runtime: `python /scripts/vercy.py query .` - Local MCP runtime: `python /mcp/server.py --dimension .` - Compatibility: `compatibility.yaml` Read this file before accessing or changing the Dimension. This file defines the local boundary and may narrow, but never silently widen, higher-level safety rules. ## Purpose and boundary A synthetic but complete public reference Dimension for testing Vercy runtime behavior. In scope: To be refined with the owner. Out of scope: Anything not explicitly placed in scope. ## Authority - Owner: public-example:vercy - Stewards: To be declared. - Trusted agents: To be declared. - Autonomous: local reading, indexing, validation, published compatible model installation, dependency resolution and typed model linking inside policy - Propose: draft models, schemas, undeclared sources and material semantic changes - Confirm: publication, access expansion, external writes, deletion, secrets and irreversible effects - Forbidden: actions outside owner authority, policy or law ## Storage and interfaces - Canonical binding: files-markdown - Interfaces: Local file access unless the owner declares another interface. - Master systems: To be declared before importing facts. - Storage decision: `bindings/storage.yaml` ## Startup and federation Read `memory.yaml`, `federation.yaml`, `context-routing.yaml`, `policies/autonomy.yaml` and `policies/model-deployment.yaml` before data. The read plane may combine authorized projections from multiple Dimensions. Every write must have exactly one owning Dimension and master system. Never copy private Personal facts into an Organization merely for convenience. When Vercy is enabled for this scope, poll due request UUIDs, resolve structural needs through the API and reconcile published compatible models without asking about storage or dependency mechanics. If a model is not ready, retain the server's `vercy-pending-model/v1` annotation on provisional data. When it is published, install it, semantically refactor that data, validate and record the migration. Ask only for new authority or an indispensable fact that cannot be safely inferred. ## Read and change Read `dimension.yaml`, `vercy.lock`, policies, routing and registries before model data. Read each model's `AGENTS.md` before its specification. Preserve identity, mastership, provenance, uncertainty, access and RFC 3339 timestamps. Never invent missing facts or mutate an immutable released specification. Runtime records under `data/objects`, `data/facts`, `data/relations` and `data/events` are immutable. Use the skill writer, its separate `.vercy/write.lock` lease and object `--expected-head` check. `vercy.lock` only pins model versions and digests; it is never a write mutex. Rebuild the replaceable SQLite index after writes and validate before reporting success. Current truth is bitemporal. Follow `policies/conflict-resolution.yaml`: explicit supersession wins, then declared authority. If equally authoritative assertions disagree, return the contested set and never guess a winner. ## Completion Report locations, model versions and digests, sources, answered questions, unknowns, validations, events and authority still required.