REE Architecture Streams and IDs (v1)

This file defines the canonical identifiers shared by:

architecture_static.mmd
architecture_typed_dataflow.mmd
episode_sequence.mmd
stream_routing.v1.yaml

Canonical component identifiers

ID	Role
`ENV.world_state`	External world state and consequences
`EXT.external_inputs`	External inputs from user/tool/sensor channels
`BOUNDARY.typed_payload_boundary`	Runtime typed ingress (`OBS`/`INS`)
`REE.shared_latent_substrate`	Shared JEPA-like latent substrate used by fast/slow regime adapters
`REE.E1_persistent_predictive_substrate`	Slow/deep regime adapter over shared substrate
`REE.E2_fast_forward_predictor`	Fast/near-horizon regime adapter over shared substrate
`REE.HippocampalSystems_rollout_and_provenance`	Explicit rollout generation and provenance binding
`REE.RC_reality_coherence_loop`	Reality-coherence conflict lane (`S5`/`RC_conflict`)
`REE.error_channel_router`	Tagged error bus with sink/write-locus gating
`REE.control_plane_router`	Control routing over stream/loop/global planes
`REE.thalamic_precision_relay`	Precision/delay relay from control plane into E1/E2/E3
`REE.neuromodulatory_state_estimator`	Neuromodulatory state estimate (precision/gain regime input)
`REE.connectome_constraint_model`	Connectome/effective-connectivity constraint feedback for routing
`REE.E3_trajectory_selector`	Commitment gating and trajectory selection
`REE.associative_meta_calibration`	Outcome-vs-decision calibration for control-plane retuning
`REE.filter_context_anchor`	Pre-commit contextual/provenance eligibility filter
`REE.filter_body_feasible`	Pre-commit homeostasis/feasibility eligibility filter
`REE.filter_motivation_ok`	Pre-commit motivational sufficiency eligibility filter
`REE.filter_hard_veto`	Pre-commit hard-veto eligibility filter
`REE.gate_cognitive_set`	Cognitive-set commit gate
`REE.gate_motivational`	Motivational commit gate
`REE.gate_motor`	Motor/action release gate
`REE.gate_probe_sweep`	Rapid disinhibitory cross-check sweep before release
`REE.action_executor`	Action execution surface
`REE.post_commit_learning`	Post-commit attribution and durable update router
`REE.verifier`	Verifier for privileged commitment paths
`REE.trusted_stores_POL_ID_CAPS`	Trusted authority stores

Stream and signal identifiers

Payload boundary types

payload.OBS
payload.INS
trusted.POL
trusted.ID
trusted.CAPS

Shared-substrate predictive interface channels

latent.shared_state
latent.slow_update_proposal
latent.fast_update_proposal
latent.prior_bundle
forward_prediction_kernel
rollout.candidates
provenance.bindings

Tagged error-router channels

error.sensory_prediction
error.action_prediction
error.precommit_sim
error.postcommit_outcome
error.precommit_sim+error.action_prediction+error.sensory_prediction
error.postcommit_outcome_replay

Control-plane signal and routing channels (MECH-004/064/065)

signal.S1_fast
signal.S1_slow
signal.S3_fast
signal.S3_slow
signal.S5_RC_conflict
control.modulator_state
route.candidate_profile
route.constraint_feedback
control.precision_bias_bundle
control.precision_thal
control.delay_bias_thal
control.Pi_ext+Pi_prop
control.Pi_int+Pi_rc+Pi_noc
control.K3+K4+K5+K10
control.P_cognitive
control.meta_recalibration
commit.inhibit
search.widen

Commit-boundary learning channels (MECH-060/061/062)

commit_boundary.tokenized_update
post_commit.outcome_trace
packet.decision_snapshot
packet.outcome_snapshot

commit_boundary.tokenized_update events should carry explicit lineage and safety-override metadata:

required: commit_id, trace_id, phase
optional lineage: parent_commit_id, supersedes_commit_id
optional supersession fields (when applicable): supersession_reason, hazard_class, override_scope

This keeps post-dispatch interruption append-only and auditable rather than allowing silent rewrite of original commit accountability.

Logging event envelope (recommended)

{
  "event_id": "evt_000001",
  "ts": "2026-02-21T00:00:00Z",
  "from": "REE.E2_fast_forward_predictor",
  "to": "REE.control_plane_router",
  "type": "signal.S1_fast",
  "phase": "pre_commit",
  "commit_id": null,
  "trace_id": "trace_abc123"
}

Naming rules

Components use dotted identifiers with stable prefixes (ENV, EXT, BOUNDARY, REE).
Edge labels use loggable stream IDs (stream.*, signal.*, payload.*, control.*, trusted.*, error.*, route.*, gate_probe.*).
The same component IDs must appear in all three Mermaid views.
New channels are added to stream_routing.v1.yaml first, then diagram files are updated.