Docs: record that sessions are strand provenance attributes, not graph nouns (cite AIΩN Paper VII)

[flyingrobots]

Context

docs/design/built-in-echo-graph-data-model.md declares the durable graph
nouns (WorldlineId — "durable lane identity", StrandId, SettlementOutcome…)
and never mentions sessions. Meanwhile an earlier design thread discussed
"writers use sessions; sessions admit intents to causal lanes; make the
session a node in the Echo graph" — and a reader who half-remembers that
thread finds an absence instead of an answer. (Exactly this happened on
2026-06-10.)

Authority

AIΩN Paper VII (April 2026, DOI 10.5281/zenodo.19751149) resolves it:

• §6.3 (Three-Tier Thinking Room): debugger-created strands are real causal
  objects whose "provenance should record the creating principal, tool or
  session origin, fork basis, and retention or revelation posture."
• §6: "a process is a strand whose live realisation appears as a shadow
  working set over shared machine history."
• Glossary B.2/B.4: lanes, worldlines, strands, braids are the history
  nouns; no Session noun exists at ontology level.

Sessions are origin/provenance attributes on strands, not lanes and not
graph nodes. The graph model's omission is conformant, not drift.

Proposal

Add a short "Sessions" note to the graph data model doc stating the
decision and citing Paper VII §6.3, so the absence reads as doctrine. If
the session-as-node idea was intended to survive post-Paper-VII, this
issue is the place to say so instead — in which case it becomes a frontier
design item, not a doc fix.

Related: graft BEARING carries the mirror tension ("session is too
transport-scoped; move toward a strand-scoped causal envelope") and graft
already records causalSessionId/strandId/warpWriterId in continuity
evidence, consistent with sessions-as-provenance.

[coderabbitai]

🔗 Related PRs

#358 - feat(core): add admission evidence boundaries [merged]
#382 - 0024 universal LE binary codec + 0025 sessions design (design-only) [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[flyingrobots]

Correction — the decision record exists and says more than this issue did.

docs/design/0025-sessions-as-causal-contexts/ (request + design + phase-2
handoff) is the recovered session thread, and it promotes Session to a
first-class durable causal-context node — principal-bound
(optic_artifact::PrincipalRef), lifecycle-gated, with an admission gate
over submissions, attribution on every intent/effect/receipt, and a
queryable lifecycle-event projection. So "session as a node in the Echo
graph" survived — with a sharp boundary this issue should quote instead of
my earlier Paper-VII-only inference:

Worldline, head, tick, strand, braid — mutation ordering /
state-lineage primitives. Session attributes, it does not reorder.

and "Session event projection (not a 'lane')". The reconciliation:
Session IS a graph noun (causal context / attribution), and is NOT a
mutation-ordering lane — which is also exactly why the built-in graph
data model (whose nouns are ordering/lineage primitives) omits it; both
docs are right, but neither cites the other.

Revised proposal for this issue: the graph-data-model doc should reference
design 0025 for the Session noun (causal-context node, non-ordering), and
0025's "Naming reclaim of Session" section is the authoritative answer to
"writers use sessions; sessions admit intents to causal lanes; session as
node." Paper VII §6.3 (session origin recorded in strand provenance)
remains compatible as the attribution view. Motivating context worth
keeping visible: 0025 exists to delete jedit's client-side
JeditWorldlineSessionPort bridge, and StrandRegistry is already
session-scoped in warp-core.