Clarify Continuum-shaped structural evidence boundary

NORTHSTAR.md

@@ -59,6 +59,15 @@ or host projection as the primary ontology. The shared boundary is witnessed
 causal history plus the Continuum-authored contract families that make that
 history admissible, observable, exportable, and importable.
 
+The critical migration distinction is:
+
+```text
+Continuum-shaped, not Continuum-native.
+```
+
+Graft may normalize structural readings into a Continuum-compatible shape.
+Only Continuum-producing runtimes may claim Continuum-native witnesshood.
+
 The stack split is:
 
 | Layer | Owner | Responsibility |
@@ -130,6 +139,8 @@ A reading should name:
   rights-limited, unavailable, obstructed, or intentionally degraded
 - **support**: witness, receipt, shell, retained artifact, provenance payload,
   or explicit absence of support
+- **evidence status**: Continuum-native when backed by a native Continuum
+  witness, or translated-substrate when adapted from a non-native substrate
 - **payload identity**: digest, version, schema, or generated artifact id
 - **typed payload**: the actual Graft-owned structural answer
 
@@ -144,7 +155,7 @@ The long-term path is:
 ```text
 Continuum shared family
   -> Wesley-generated artifacts
-  -> Echo or git-warp runtime publication
+  -> Echo runtime publication or translated git-warp adapter evidence
   -> ObservationRequest / ObserverPlan
   -> ReadingEnvelope-backed runtime result
   -> Graft structural payload
@@ -154,6 +165,8 @@ Continuum shared family
 In practice:
 
 - `git-warp` remains the right adapter for cold, Git-backed committed history.
+  Until it publishes Continuum boundary artifacts directly, its evidence is
+  translated substrate evidence, not a Continuum-native witness.
 - Echo becomes the right adapter for hot, live, frontier-oriented histories.
 - Continuum owns shared runtime-boundary nouns such as `ObserverPlan`,
   `ObservationRequest`, `ReadingEnvelope`, `WitnessedSuffixShell`,
@@ -223,7 +236,8 @@ The move is:
 git-warp-shaped implementation
   -> substrate-neutral structural reading port
   -> git-warp adapter behind the port
-  -> Continuum runtime-boundary fixtures
+  -> translated-substrate evidence with nativeContinuumWitness: false
+  -> Continuum-native runtime-boundary fixtures
   -> Echo/live-frontier adapter
   -> generated shared artifacts where the family is no longer Graft-local
 ```
@@ -236,9 +250,9 @@ architecture honest.
 1. Ship `v0.8.0` as the final git-warp-first Review Truth release.
 2. Introduce a substrate-neutral structural reading port in Graft.
 3. Keep the existing git-warp committed-history implementation behind that
-   port.
-4. Add fixture-backed Continuum runtime-boundary coverage for `ObserverPlan`,
-   `ObservationRequest`, and `ReadingEnvelope`.
+   port and mark its evidence as translated/non-Continuum-native.
+4. Add fixture-backed or Echo-backed Continuum runtime-boundary coverage for
+   `ObserverPlan`, `ObservationRequest`, `ReadingEnvelope`, and evidence status.
 5. Prove one Echo or `jedit` live-frontier structural reading.
 6. Shape Graft-owned structural payloads so they can be wrapped in
    `ReadingEnvelope` posture without making Continuum own premature app nouns.
@@ -258,7 +272,7 @@ We know the north star is becoming real when these things are boring:
 - A PR review over committed Git history and a live editor reading over Echo
   both return evidence-bearing structural readings.
 - API, CLI, and MCP callers receive the same basis, freshness, residual,
-  support, and payload identity posture.
+  support, evidence-status, and payload identity posture.
 - `warp-ttd` can show a Graft structural reading alongside runtime receipts
   without hand-normalizing host-specific stories.
 - Continuum owns the shared families; Wesley generates the artifacts; runtimes

docs/BEARING.md

@@ -11,14 +11,24 @@ timeline
 
 ## Active Gravity
 
-### 1. Entrypoint Convergence
+### 1. Continuum-Shaped Structural Reading Port
+- Defining `StructuralReadingPort` as the only Graft-facing structural read
+  boundary.
+- Keeping the current git-warp committed-history behavior behind that port
+  while marking its evidence as translated/non-Continuum-native.
+- Proving at least one fixture-backed or Echo-backed Continuum-native reading
+  branch without making Graft a Continuum semantic owner.
+- Routing API, CLI, MCP, and rendering paths through normalized Graft structural
+  payloads rather than substrate-specific facts.
+
+### 2. Entrypoint Convergence
 - Formalizing API, CLI, and MCP as equal first-class entry points.
 - Extracting application services so those three surfaces stop owning
   business flow.
 - Establishing baseline capability posture and parity expectations
   before more surface growth lands.
 
-### 2. WARP Ontology & Causal Collapse
+### 3. WARP Ontology & Causal Collapse
 - Explicit definition of session, strand, and checkout epoch.
 - Implementation of strand-aware causal collapse (admission of speculative work into canonical history).
 - Strengthening of symbol identity and rename continuity for precise slicing.
@@ -27,12 +37,12 @@ timeline
   WARP reference and precision paths have been mitigated; medium-severity
   local-history and newer structural-metric reads remain tracked.
 
-### 3. Multi-Repo Coordination
+### 4. Multi-Repo Coordination
 - Refinement of the Shared Daemon trust boundaries.
 - System-wide resource pressure and fairness summaries across multiple repos.
 - Authorization-filtered multi-repo overview surfaces.
 
-### 4. Agentic Observability
+### 5. Agentic Observability
 - Implementation of the Deterministic Scenario Replay pipeline.
 - Machine-readable between-commit activity views for agents and humans.
 
@@ -58,25 +68,28 @@ enhance, governed-edit, path-boundary, and Dockerized validation spine.
 `v0.7.1` followed with npm package hygiene: built `dist/` runtime
 artifacts, no published `src/`, development-only `tsx`, and release
 publish guards.
+`v0.8.0` shipped Review Truth: structural PR summaries, provenance
+signals, symbol history, removed-symbol evidence, structural
+test-reference signals, review readiness, and broader language/config
+parsing.
 
 ## Next Target
 
-The immediate focus is **v0.8.0 release prep**, not more feature work.
+The immediate focus is the **Continuum-shaped structural reading port**, not a
+runtime rewrite.
 
-1. Keep `main` release-clean after `v0.7.1`.
-2. Land `v0.8.0` as a **Review Truth** release: structural PR review
-   summaries, bounded provenance hints, symbol history, removed-symbol
-   evidence, structural test-reference signals, automated review
-   readiness, and the first language-breadth stack for Rust, GraphQL,
-   Python, Go, JSON, TOML, and YAML.
-3. Treat remaining `docs/method/backlog/v0.8.0/` cards as follow-up
-   candidates unless release validation exposes a direct blocker.
-4. Do not add METHOD-specific backlog/status features to Graft. METHOD
+1. Keep `main` clean after the `v0.8.0` release.
+2. Introduce `StructuralReadingPort` as Graft's single structural read boundary.
+3. Put git-warp committed-history reads behind the port and label them as
+   translated/non-Continuum-native evidence.
+4. Prove the Continuum-native branch with fixture-backed or Echo-backed
+   evidence before any adapter claims native witnesshood.
+5. Do not add METHOD-specific backlog/status features to Graft. METHOD
    backlog lanes, cards, retros, dependency DAGs, and release truth
    surfaces belong in Method MCP / Method CLI.
-5. Keep `WARP_lsp-enrichment` and `CORE_migrate-to-slice-first-reads`
-   out of this release. LSP enrichment remains valid optional scope;
+6. Keep `WARP_lsp-enrichment` and `CORE_migrate-to-slice-first-reads`
+   out of this slice. LSP enrichment remains valid optional scope;
    slice-first reads remain externally blocked until git-warp observer
    geometry APIs land.
-6. Treat daemon live refresh and daemon control-plane actions as a
-   separate daemon-operator lane, not part of the v0.8.0 cut.
+7. Treat daemon live refresh and daemon control-plane actions as a
+   separate daemon-operator lane, not part of this slice.

docs/method/backlog/up-next/CORE_continuum-structural-reading-port.md

@@ -12,24 +12,29 @@ requirements:
   - "Existing Graft WARP/git-warp review, symbol-history, dead-symbol, provenance-hint, and structural-test-reference surfaces"
   - "Existing API / CLI / MCP capability posture"
 acceptance_criteria:
-  - "Graft exposes one substrate-neutral structural reading port that does not import Echo, git-warp, or warp-ttd concrete runtime types"
+  - "StructuralReadingPort exists as the only Graft-facing structural read boundary"
   - "The existing git-warp-backed committed-history implementation sits behind that port without changing v0.8.0 review behavior"
-  - "At least one fixture-backed Continuum runtime-boundary reading drives a Graft structural projection test"
-  - "Structural reading results name basis, freshness, residual posture, witness or shell reference, and payload identity"
+  - "git-warp readings are explicitly marked translated/non-Continuum-native"
+  - "No git-warp commit/range evidence is modeled as a Continuum witness"
+  - "At least one fixture-backed or Echo-backed test proves the Continuum-native evidence branch"
+  - "API, CLI, MCP, and rendering paths consume the normalized Graft structural payload instead of substrate-specific facts"
   - "The design names Graft's proposed Continuum registry role without making Graft the semantic owner of shared families"
 ---
 
 # Continuum structural reading port
 
 ## Hill
 
-Make Graft's structural memory boundary Continuum-shaped instead of
-`git-warp`-shaped.
+Make Graft's structural memory boundary Continuum-shaped, not
+Continuum-native.
 
 By the end of this slice, Graft should have one explicit structural reading
-port that can support the current cold committed-history implementation while
-preparing for Echo/live-frontier readings through Continuum-authored boundary
-families.
+port that accepts Continuum-shaped boundary concepts without requiring every
+substrate to publish native Continuum artifacts.
+
+The first implementation must support the current cold committed-history
+implementation while preparing for Echo/live-frontier readings through
+Continuum-authored boundary families.
 
 ## Why
 
@@ -57,7 +62,9 @@ The stack direction is now clearer:
 - Graft is the structural-code observer and review engine.
 
 Graft should therefore consume and emit structural readings over witnessed
-causal history, not treat a materialized graph as the primary ontology.
+causal history where native evidence exists, and over explicitly translated
+substrate evidence where it does not. A Continuum-shaped payload is not the
+same thing as a Continuum-native witness.
 
 ## Proposed Continuum role
 
@@ -72,16 +79,48 @@ the same interchange semantics. If a structural payload family becomes shared
 across tools, promote it into Continuum through the normal family promotion
 rules instead of copying local DTOs.
 
+## Evidence rule
+
+Graft may normalize readings into Continuum-compatible shape. Only
+Continuum-producing runtimes may claim Continuum-native witnesshood.
+
+The port should model evidentiary status directly:
+
+```ts
+type StructuralReadingEvidence =
+  | ContinuumNativeEvidence
+  | TranslatedSubstrateEvidence;
+
+type ContinuumNativeEvidence = {
+  kind: "continuum-native";
+  envelope: ReadingEnvelope;
+  witness?: WitnessedSuffixShell;
+};
+
+type TranslatedSubstrateEvidence = {
+  kind: "translated-substrate";
+  substrate: "git-warp";
+  basis: GitWarpCommittedBasis;
+  evidence: GitWarpEvidence;
+  nativeContinuumWitness: false;
+};
+```
+
+The ugly `nativeContinuumWitness: false` marker is intentional. It prevents a
+Git commit/range, git-warp graph fact, or adapter-produced compatibility object
+from being laundered into a Continuum witness.
+
 ## Expected shape
 
 The first port should describe Graft's needs, not a generic graph database:
 
-- source/basis identity
+- basis identity
 - observation request identity
 - structural reading kind
 - current/stale/incomparable freshness
 - complete/partial/plural/budget-limited/rights-limited/unavailable posture
-- witness, receipt, shell, or hologram reference when present
+- evidence status: Continuum-native or translated-substrate
+- witness, receipt, shell, or hologram reference only when genuinely present
 - payload digest or identity
 - typed payload for Graft-owned structural facts
 
@@ -91,6 +130,7 @@ The current git-warp-backed implementation should become one adapter:
 Graft review / symbol history / dead symbols / test references
   -> StructuralReadingPort
   -> git-warp committed-history adapter
+  -> translated-substrate evidence
 ```
 
 A future Echo/live-frontier implementation should become another adapter:
@@ -100,6 +140,7 @@ Graft live frontier structural projection
   -> StructuralReadingPort
   -> Continuum runtime-boundary family
   -> Echo or jedit head/frontier basis
+  -> continuum-native evidence
 ```
 
 `warp-ttd` should consume Graft structural readings as observer artifacts. It
@@ -110,20 +151,28 @@ hand-normalized against Echo and `git-warp`.
 
 1. Define the structural reading port in `src/ports/` or the existing core
    boundary that best matches the local architecture.
-2. Model only the minimum basis, request, residual posture, witness reference,
-   and payload identity needed by current Review Truth behavior.
+2. Define `StructuralReadingEvidence` as a union of Continuum-native evidence
+   and translated substrate evidence.
 3. Move current git-warp-facing review/symbol/dead-symbol/test-reference reads
    behind the port without changing their public output contracts.
-4. Add a fixture-backed Continuum runtime-boundary test using the existing
-   GraphQL fixture posture as input evidence, then replace the fixture with
-   generated artifacts when the stack is ready.
-5. Document the Graft/Continuum/warp-ttd split in the design packet and keep
+4. Mark every git-warp committed-history reading as `translated-substrate` with
+   `nativeContinuumWitness: false`.
+5. Add a fixture-backed or Echo-backed Continuum runtime-boundary test proving
+   the Continuum-native branch, then replace fixtures with generated artifacts
+   when the stack is ready.
+6. Route API, CLI, MCP, and rendering paths through the normalized Graft
+   structural payload instead of substrate-specific facts.
+7. Document the Graft/Continuum/warp-ttd split in the design packet and keep
    [NORTHSTAR.md](../../../../NORTHSTAR.md) aligned.
 
 ## Non-goals
 
 - Do not rewrite `v0.8.0` release scope.
 - Do not replace git-warp in one step.
+- Do not require git-warp to publish native Continuum artifacts before it can
+  back Graft's structural port.
+- Do not model git-warp commit/range evidence as a Continuum witness.
+- Do not let "Continuum-shaped" imply "Continuum-native".
 - Do not add a direct Echo dependency before the port boundary exists.
 - Do not make Graft the owner of Continuum runtime-boundary nouns.
 - Do not make `warp-ttd` the structural-review engine.