Mission is built around strict ownership:

  • Entity classes own domain behavior and invariants.
  • Entity schemas own validated payload, storage, and data shapes.
  • Entity contracts expose daemon-readable method metadata.
  • The daemon owns runtime state, command dispatch, and agent coordination.
  • Open Mission is a surface over daemon-owned state.
  • Adapters translate external systems into Mission concepts.

Generic Entity infrastructure is child-independent. It must not import concrete Entity contracts, concrete Entity classes, daemon registries, runtime services, adapters, or provider implementations. Concrete contract catalogues and daemon capability wiring belong in daemon-owned dispatch modules. See ADR-0001.07 before changing Entity infrastructure.

Read System Context, Semantic Model, Entity Reference, and Entity Command Surface first. ADR-0001.03 is the key architectural decision behind the current OOD model.

For daemon-supervised live frontend work served through the Open Mission host, read Impeccable Live Supervision. It defines the owner-id-only live-session model, daemon runtime-supervision ownership, host proxy behavior, and operator usage expectations.

Temporary working specs are allowed while a refactor is in flight. For Agent execution structured interaction, use Agent Execution Structured Interaction Spec as the current implementation reference until its durable decisions are folded back into CONTEXT.md and accepted ADRs. For the daemon-owned MCP signal transport, use Mission MCP Server Spec as the current realization blueprint for ADR-0006.06.

For the current journal-ledger refactor, use Agent Execution Journal Ledger Spec as the temporary implementation reference for the typed journal family split and replay contract tightening.

For the next AgentExecution persistence step, read Agent Execution Interaction Journal PRD and Agent Execution Interaction Journal Spec. These documents define the canonical semantic interaction journal that separates AgentExecution replay truth from terminal recordings and Mission workflow events.

For the Mission-native code intelligence slice, read Code Intelligence PRD and Code Intelligence Spec. These temporary documents define the GitNexus-inspired, SurrealDB-backed code graph and the open-mission-mcp semantic operations that expose it to scoped Agent executions.

For the next code-intelligence graph simplification step, read Code Intelligence Node Edge Graph PRD and Code Intelligence Node Edge Graph Spec. These temporary documents define the minimal persisted graph shape built around code_object, code_relation, and explicit snapshotId ownership.

For the normalized Entity storage refactor, read Entity OGM Storage/View PRD and Entity OGM Storage/View Spec. These temporary documents define the /flying-pillow-style OGM-backed split between canonical storage rows and hydrated Entity views, plus the relation-aware registry and SurrealDB direction needed for the general Entity-model refactor.

For the concrete realization of that storage model in current code, read Entity OGM Hydration Pattern. It explains the general Entity class plus storage-schema plus hydrated-schema pattern, and uses Terminal as the clearest working example.

For the structured-first Agent interaction split, read Agent Interaction Structured-First PRD and Agent Interaction Structured-First Spec. These temporary documents define the structured Mission control lane, optional terminal capability, slash-command taxonomy, and migration path behind ADR-0006.10.

Table of contents