V3 convergence wiring: tenant-carve certification, RungElevator, P6 wave probe, seam-list plan#626
Merged
Merged
Conversation
…s the reserved root slot Surfaced when merged OGAR #145 + lance-graph #624 met: OGAR shipped osint_system at 0x0700 (CC==0x00, the reserved domain-root slot per canon), which the mirror's in-tree zero-slot test forbids — so the parallel-mirror is blocked on a remap decision (Option A: move to 0x0702 + q2 update; Option B recommended: 0x0700 is the OSINT domain-root/default class, drop from the concept CODEBOOK, mirror only osint_person 0x0701). lance-graph main CI is green (fuse is in the excluded lance-graph-ogar); consumers vendoring lance-graph-ogar are the exposure. Board: ISSUES ISS-OSINT-SYSTEM-ROOT-SLOT-VIOLATION (blocks ISS-OGAR-OSINT-MIRROR-PENDING). Co-Authored-By: Claude <noreply@anthropic.com>
Frozen concept (lo u16) = the AR class identity (a class is just a class). The non-frozen 0x1000 app prefix is the render side that adapts the askama SoA view: class -> ClassView -> bitmask (implicit focus of attention over SoA columns) -> askama template (Redmine-ERB over AR). No semiring on the view path. The osint_system 0x0700->0x0702 fix is a frozen-side lo-u16 slot correction, orthogonal to this render doctrine. Board: EPIPHANIES E-RENDER-IS-CLASS-BITMASK-ASKAMA-NOT-SEMIRING. Co-Authored-By: Claude <noreply@anthropic.com>
… + unblock resolve via local ndarray patch Phase-2 seam of the V3 substrate: the AriGraph-shaped SoA tenants already exist (ValueSchema::Cognitive = the 7 hot lanes the MailboxSoaView columns expose); what was missing was their certification. New probe osint_v3_cognitive_tenant_carve_field_isolation_matrix extends the I-LEGACY mandatory matrix from Kanban-only to the whole Cognitive carve on a registry-minted OSINT-V3 row (mint_for dispatch, never a hardcoded constructor): per-tenant lane flip changes zero bytes outside the lane, key+edges untouched, EntityType lane carries the canon 0x0700 concept (the 0x1000 gen-marker never leaks into the entity discriminator), and the typed accessors decode the same certified slab. Build unblock: [patch.crates-io] ndarray git-URL -> local sibling path. The git form re-fetched the fork + its burn submodule on every resolve; AdaWorldAPI/burn is outside the session repo scope (403; gitlink 9b2b671 unfetchable), deadlocking scoped/offline sessions. The patch was [[patch.unused]] in the lock either way (TD-NDARRAY-PATCH-0_16), so the resolved graph is unchanged - only the fetch deadlock is gone. Same fork, local source, per the P0 fork doctrine. facet_schema.rs: fmt normalization only. Gates: 762 lib tests green (guid-v2-tail,guid-v3-tail), 749 default, cargo fmt clean, clippy -D warnings clean. Board: EPIPHANIES E-V3-TENANTS-ALREADY-EXIST-WIRE-DONT-INVENT (same commit). Co-Authored-By: Claude <noreply@anthropic.com>
…overage complete Cold half of the Phase-1 V3 set: FMA-V3 and CPIC-V3 both resolve to ValueSchema::Compressed (Fingerprint/HelixResidue/TurbovecResidue/ EntityType). Same certification as the OSINT-V3/Cognitive matrix: pin the 4-lane carve (and the 6 absent hot tenants), registry-dispatched mint_for, per-tenant lane-flip isolation via the shared assert_value_lane_isolation helper (Cognitive matrix refactored onto it), key+edges untouched, EntityType discriminator carries the canon lo-u16 concept (Anatomy 0x0A01 / Genetics 0x0E00 — the 0x1000 gen-marker never leaks). With this, every carve a Phase-1 V3 class materialises is matrix- covered; Phase-2 tenant readings proceed over certified lanes. Gates: 763 lib tests green (guid-v2-tail,guid-v3-tail), 749 default, cargo fmt clean, clippy -D warnings clean. Board: EPIPHANIES E-V3-TENANTS-ALREADY-EXIST-WIRE-DONT-INVENT Status line updated (both probes named). Co-Authored-By: Claude <noreply@anthropic.com>
…) + stale-brief correction q2's cpic::NodeGuid mirror verified by read: V1 byte-layout parity TRUE, but V3 (part_of:is_a) tiles ride a V1 u24 tail under unregistered domain 0x0C — three divergences from the registered CPIC-V3 read-mode (0x1000_0E00, V3 tail). Record-only; resolution is an operator decision (q2 is push-gated; cross-repo blockers are never silently fixed). Also corrects the stale soa-value-tenant-migration-v1.md §2.5 blocker: new_v2 exists (7 groups, guid-v2-tail, matrix-tested) and q2 has no new_v2 call site today. Co-Authored-By: Claude <noreply@anthropic.com>
…ated-grep error owned) osint-bake/src/lib.rs:606,745 mint via NodeGuid::new_v2 against the REAL contract import (no mirror); only cpic carries a local mirror. Appended dated correction per the append-only discipline. Also records the known tail-shape pending: OSINT rows mint a V2 tail for legacy CLASSID_OSINT (registered V1) while the FMA bins already use the mint_for dispatch; CLASSID_OSINT_V3 exists for that migration. Co-Authored-By: Claude <noreply@anthropic.com>
…graph 124/124 green in-sandbox
The module's declared caveat ('lance-graph core does NOT build in the
offline sandbox') is obsolete: with crates.io in the proxy allow-list,
the root [patch.crates-io] ndarray on the local sibling path, and protoc
installed, cargo check -p lance-graph finishes clean and
cargo test -p lance-graph --lib graph::arigraph passes 124/124 —
including markov_soa's 4 wave-projector tests. The wire is probed
end-to-end: V3 classid -> registry read-mode -> matrix-certified tenant
carves -> AriGraph SoA wave.
Board: EPIPHANIES E-V3-TENANTS-ALREADY-EXIST-WIRE-DONT-INVENT Status
addendum (same commit).
Co-Authored-By: Claude <noreply@anthropic.com>
… sweep Logs the session arc (patch unblock, both carve matrices, core in-sandbox first build, markov_soa STATUS clear) and the Sonnet report-only sweep: core 925/925, planner 204/204, supervisor green, zero stale offline-status comments. Co-Authored-By: Claude <noreply@anthropic.com>
…seam-list plan
D-VCW-1a — the rung ladder as dispatch policy over certified mask algebra:
RungLevel::{from_u8, elevate, de_elevate, pearl_level, causal_mask_bits}
(0-2 observe / 3-5 intervene / 6-9 counterfactual — the enum names its own
Pearl boundary at Counterfactual=6; PO/SPO masks P3-certified, O-for-L1 a
labeled convention pending probe) + RungElevator (sustained-BLOCK elevates,
sustained-FLOW relaxes to the dispatched base, HOLD resets streaks;
threshold 2 hand-tuned, disclosed). Converged with the EXISTING
escalation::rung_delta felt-parse hint via apply_delta — one ladder, two
signal sources (gate streaks = System-2, qualia delta = System-1), no
parallel type invented. 7 new tests; contract 755 green; clippy clean.
D-VCW-2 — P6: the AriGraph wave's injected distance is the SAME certified
256x256 palette read as the P1-P3 particle chain (self-match exactly 1.0;
hand-computed table arithmetic == best_guess_match output). markov_soa
6/6 green. (Sonnet grindwork per the session model split.)
D-VCW-6 — worker Rule 7: negative-existence claims require an
exhaustive-search declaration (born from this session's truncated-grep
false claim, corrected on the board).
Plan: .claude/plans/v3-convergence-wiring-v1.md (the seam list: every
deliverable a probe or a wiring of existing types; two-SoA-worlds doctrine;
P7 render-probe spec for q2; one-row registry direction; D7 deferred).
Board (same commit): INTEGRATION_PLANS prepend, STATUS_BOARD D-VCW rows,
EPIPHANIES E-RUNG-LADDER-IS-DISPATCH-POLICY-OVER-CERTIFIED-MASKS,
ISSUES ISS-Q2-CASCADE3-NIBBLE-ANCESTRY + Option-B ground-truth addendum
on ISS-OSINT-SYSTEM-ROOT-SLOT-VIOLATION (two id spaces aliasing in the
lo u16 — record-only, operator's decision).
Co-Authored-By: Claude <noreply@anthropic.com>
…tx.rung proxy retired The sustained-BLOCK elevation policy goes live in the driver: - ShaderDriver gains a persistent RwLock<RungElevator> (same updated-every-cycle pattern as awareness) — a per-call-local elevator could never accumulate a streak across dispatches, since each run() is one cycle by this module's own definition. - A dispatch requesting a different base rung resets the elevator to that base: streaks never leak across unrelated dispatch contexts. - The cycle's already-decided gate feeds elevator.on_gate() AFTER the gate decision — provenance never alters the gate (invariant kept). - materialize_provenance takes the live rung parameter; the documented 'ctx.rung <- 1 (proxy)' is retired. Rung now reaches ThoughtCtx tier selection (materialize.rs rung >= 7 / >= 4 gating). - wire.rs + grpc.rs: identical hand-rolled 10-arm u8->RungLevel matches deduped through the contract's RungLevel::from_u8. Tests: driver 100/100 green, incl. 2 new — sustained-BLOCK elevation across real dispatch() calls (deterministic BLOCK via empty row window) + rung load-bearing in tactic selection (honesty note in-test: tier is a +1 tie-weight; inequality asserted for the empirically-differentiating input, rung 1 -> tactic 17 vs rung 9 -> tactic 3, not claimed universal). Contract regression 755 green; fmt clean; driver-own lints clean. Implementation: Sonnet grindwork agent (killed mid-test by a worker restart), reviewed + finished on the main thread. Board (same commit): STATUS_BOARD D-VCW-1b Shipped, plan D1b/D2 status, AGENT_LOG entry for both agents, TECH_DEBT TD-DEPRECATED-ACCESSORS-BLOCK-DEP-CLIPPY (pre-existing ontology/planner deprecations fail any dependent's clippy -D warnings run). Co-Authored-By: Claude <noreply@anthropic.com>
|
Warning Review limit reached
Next review available in: 49 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (4)
📝 WalkthroughWalkthroughIntroduces a ChangesRung Elevation and V3 Substrate Verification
Board, Plan, and Knowledge Documentation
Estimated code review effort: 3 (Moderate) | ~30 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c9588b2f1b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| 8 => RungLevel::Recursive, | ||
| _ => RungLevel::Transcendent, | ||
| }; | ||
| let rung = RungLevel::from_u8(req.rung as u8); |
There was a problem hiding this comment.
For gRPC callers, DispatchRequest.rung is a uint32 (proto/shader.proto:46), but this narrows it to u8 before calling RungLevel::from_u8. Values above 255 therefore wrap instead of saturating (for example, 256 becomes Surface and 257 becomes Shallow), whereas the old match treated any out-of-range value as Transcendent. Invalid client input can silently run at the shallowest depth rather than the safest maximum rung; clamp or validate while the value is still u32.
Useful? React with 👍 / 👎.
Owner
Author
There was a problem hiding this comment.
Fixed in 04712d4 — rung_from_wire_u32 saturates while still u32 (u8::try_from(v).unwrap_or(u8::MAX) → from_u8), restoring the pre-dedup _ => Transcendent behavior, with a regression test covering the exact wrap cases (10, 256, 257, u32::MAX). Verifying it also required repairing pre-existing --features grpc compile breakage (missing Pillar-7 ShaderDispatch fields in the wire/grpc initializers + stale ContextChain::new(8) test calls) — broken on the merge-base too, confirmed by stash-check; lab-feature suite now green at 186 lib tests.
Generated by Claude Code
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.claude/board/ISSUES.md:
- Around line 51-70: The note contains contradictory statements about q2
`new_v2` usage: first it says no call site exists, then it cites
`osint-bake/src/lib.rs` as calling `NodeGuid::new_v2`. Rewrite the affected
brief in `.claude/board/ISSUES.md` to a single consistent claim, using the
actual `osint-bake` call sites and the `NodeGuid::new_v2` / `classid_read_mode`
symbols to state the real migration status without the stale “does not exist”
wording.
In @.claude/plans/v3-convergence-wiring-v1.md:
- Around line 111-115: The osint `0x0700` reconciliation note uses an invalid
“Option C” reference that is not defined by the issue doc. Update the wording in
the convergence plan section so the reference matches the actual
`ISS-OSINT-SYSTEM-ROOT-SLOT-VIOLATION` options (A/B plus the addendum) by
renaming or removing the `Option C` label, keeping the rest of the `classid
human-readable reorder` note unchanged.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: cdd6cfe4-d50f-45ae-9aab-2f947456ea17
⛔ Files ignored due to path filters (1)
Cargo.lockis excluded by!**/*.lock
📒 Files selected for processing (16)
.claude/board/AGENT_LOG.md.claude/board/EPIPHANIES.md.claude/board/INTEGRATION_PLANS.md.claude/board/ISSUES.md.claude/board/STATUS_BOARD.md.claude/board/TECH_DEBT.md.claude/knowledge/autoattended-multiagent-pattern.md.claude/plans/v3-convergence-wiring-v1.mdCargo.tomlcrates/cognitive-shader-driver/src/driver.rscrates/cognitive-shader-driver/src/grpc.rscrates/cognitive-shader-driver/src/wire.rscrates/lance-graph-contract/src/canonical_node.rscrates/lance-graph-contract/src/cognitive_shader.rscrates/lance-graph-contract/src/facet_schema.rscrates/lance-graph/src/graph/arigraph/markov_soa.rs
…e-existing lab-feature breakage codex P2: DispatchRequest.rung is a proto uint32; 'as u8' truncation made 256 wrap to Surface (the shallowest depth) where the pre-dedup match saturated any out-of-range value to Transcendent. Fixed via rung_from_wire_u32 (u8::try_from().unwrap_or(u8::MAX) -> from_u8) + regression test covering the exact wrap cases (256, 257, u32::MAX). wire.rs is unaffected (its rung field is already u8). Verifying the fix required repairing pre-existing --features grpc breakage (broken on the merge-base too, verified by stash-check): - wire.rs + grpc.rs ShaderDispatch initializers missing the Pillar-7 merge_override / alpha_saturation_override fields -> None defaults (the wire DTO / proto do not carry them). - cypher_bridge.rs tests calling ContextChain::new(8) against the contract's zero-arg new(). Lab-feature suite now green: 186 lib tests incl. the new saturation test; default suite still 100/100. CodeRabbit (2 findings): plan 'Option C' label corrected to 'the dated ADDENDUM strengthening Option B' (the issue defines only A/B). The ISSUES.md rewrite request is declined on-thread: board files are append-only ledgers — corrections append as dated entries, never rewrite prior text. Co-Authored-By: Claude <noreply@anthropic.com>
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jul 2, 2026
…ion root-slot issue — RESOLVED -> INTERIM-RESOLVED The rebase onto main (#626) brought in ISS-OSINT-SYSTEM-ROOT-SLOT- VIOLATION, filed by a parallel session, which holds the osint_system 0x0700 zero-concept-slot question as an operator decision (Option A remap vs Option B drop-from-concept-codebook, with the two-id-spaces ADDENDUM favouring B). This branch's SHIPPED_ROOT_SLOT_CONCEPTS carve-out was applied before seeing that issue — it stands as the interim that keeps every consumer compiling with the fuse armed, and explicitly does NOT pre-empt the A/B decision; the Status line now records how each option unwinds the carve-out.
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jul 2, 2026
… vocabulary rows) OGAR #146 executed the operator ruling: both #145 OSINT mints removed (low byte is APPID space domain-wise), ALL back to 65 == the mirror's 65 — the COUNT_FUSE balances with zero mirror-side changes. This branch's earlier +2-rows + zero-slot carve-out commits were dropped during the rebase onto main (#626/#627) as superseded by that ruling. Verified: contract ogar_codebook 8/8 + lance-graph-ogar standalone green against a0c7936 (compile-time COUNT_FUSE holds).
This was referenced Jul 2, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The organizing finding
Wire, don't invent — every layer of the V3 substrate already contains its own solution; the gaps were unwired seams, not missing machinery (
EPIPHANIES.mdE-V3-TENANTS-ALREADY-EXIST-WIRE-DONT-INVENT). Every deliverable in this PR is either a probe or a wiring of EXISTING types; §0 anti-invention holds throughout (no newValueSchemavariants, no new carriers, no parallel registries).Full seam list + doctrine + queued work:
.claude/plans/v3-convergence-wiring-v1.md.What ships
Phase-2 tenant certification (D-VCW carve matrices)
osint_v3_cognitive_tenant_carve_field_isolation_matrix— the I-LEGACY mandatory field-isolation matrix, extended from Kanban-only to the WHOLE Cognitive carve (the AriGraph-hot tenants: Meta/Qualia/Fingerprint/Energy/Plasticity/EntityType/Kanban) on a registry-dispatchedmint_forOSINT-V3 row. EntityType lane carries the canon0x0700concept — the0x1000gen-marker never leaks into the entity discriminator.fma_cpic_v3_compressed_tenant_carve_field_isolation_matrix— same certification for the cold Compressed carve (FMA-V3 + CPIC-V3). Every carve a Phase-1 V3 class materialises is now matrix-covered.Rung ladder goes live (D-VCW-1a/1b) — zero new math
RungLevel::{from_u8, elevate, de_elevate, pearl_level, causal_mask_bits}: rungs 0–2 observe (Pearl L1), 3–5 intervene (L2 →PO = 0b011, P3-probe-certified), 6–9 counterfactual (L3 →SPO = 0b111, P3-probe-certified; the enum names its own boundary atCounterfactual = 6). L1 →O = 0b001is a labeled convention pending its own probe.RungElevator— "elevates on sustained BLOCK" (the documented intent onShaderDispatch::rung) as a pure zero-dep policy overGateDecision: sustained BLOCK elevates, sustained FLOW relaxes toward the dispatched base, HOLD resets streaks. Threshold 2 hand-tuned (disclosed perI-NOISE-FLOOR-JIRAK). Converged with the existingescalation::rung_deltaviaapply_delta— one ladder, two signal sources (gate streaks = System-2, felt-parse qualia = System-1), no parallel type.RwLock<RungElevator>onShaderDriver(base-change reset; gate fed post-decision so provenance never alters the gate); thectx.rung = 1proxy is retired — rung now reachesThoughtCtxtier selection.wire.rs/grpc.rshand-rolled u8→rung matches deduped throughfrom_u8.P6 — the wave joins the certified metric (D-VCW-2)
markov_soa::best_guess_matchdriven by a real zero-diagonal 256×256 palette table: self-match exactly 1.0; hand-computed table arithmetic equals the function's output. The AriGraph wave provably reads the SAME palette metric the particle chain (P1–P3) certified — the table is the join object (dependency flows AriGraph→sensor, never reverse).Build unblock (all sessions benefit)
[patch.crates-io] ndarraygit-URL → local sibling path: the git form re-fetched the fork + itsburnsubmodule on every resolve; burn is outside the session repo scope (403, gitlink unfetchable), deadlocking scoped/offline sessions. The patch was[[patch.unused]]in the lock either way — resolution unchanged, only the deadlock gone. Same fork, local source, per the P0 fork doctrine.markov_soa's "unverified-offline" STATUS is cleared.Verification
--libTD-DEPRECATED-ACCESSORS-BLOCK-DEP-CLIPPY)Board hygiene (same commits)
EPIPHANIES ×2, ISSUES ×2 (+ the two-id-spaces Option-B strengthening on the osint
0x0700question — record-only, operator's decision), TECH_DEBT ×1, STATUS_BOARD D-VCW rows, INTEGRATION_PLANS prepend, AGENT_LOG ×2, knowledge Rule 7 (negative-existence claims require an exhaustive-search declaration — born from this session's own corrected truncated-grep error).Explicitly NOT in this PR
OGAR untouched (osint
0x0700reconciliation awaits the operator); q2 untouched (P7 render probe + cascade3 nibble falsifier specs are ready, queued behind q2's push gate); the classid human-readable reorder stays DEFERRED-by-design.🤖 Generated with Claude Code
https://claude.ai/code/session_01MLBnPuScZy6w9di2QEjsXM
Generated by Claude Code
Summary by CodeRabbit
New Features
Bug Fixes
Tests
Documentation