Two UX wrinkles surfaced while exercising the CORE-Logos Studio:
- DAG viewport: tall, few-layer graphs (e.g. the 408x1890, 2-layer/55-node
alignment fan) were squished by viewBox-meet into an unreadable sliver. Render
at natural height (scale ~1) inside a bounded, scrollable box; small graphs are
unaffected (>= the caller height). Golden layout unchanged (layoutDag untouched).
- Click-to-copy: new shared useCopyToClipboard hook gives every copy affordance a
tooltip + transient confirmation. StatusFooter SHA now confirms 'Copied' (was a
silent copy that read as 'does nothing'); 'Read Only' is labeled a non-interactive
status (not a toggle); checkpoint-revision gains an explanatory tooltip.
DigestBadge + MetadataTable copy buttons route through the hook + gain titles.
Full workbench-ui vitest: 520 passed (+4), clean exit; tsc + build clean.
The trilingual resonance centerpiece, read-only over GET /logos/packs/{id}/alignment:
- deterministic DagViewer (he -> grc -> en longest-path) + selectable edge list;
alignmentToDag is a pure projection pinned by a golden-file layout test
- invalid alignment targets surfaced honestly (the undeclared en-collapse-*
anchors render a warning, never smoothed over)
- logos_alignment_edge evidence subject (logos_alignment_edge:<packId>/<edgeId>)
with RightInspector projection + EvidenceChainRail derivation
- holonomy stays absent: no tab, no proof card, no success element (guard test)
Full workbench-ui vitest: 516 passed, clean exit; build clean.
Read-only contents tabs over GET /logos/packs/{id}/contents:
- Lexicon (VirtualizedList + SearchInput + status/domain filters; dangling-link
flag cross-referenced from the safety report, never recomputed)
- Glosses (linked entry ids) and Morphology (operator chain rendered in schema
order — root/prefix/stem/suffix never re-sorted)
- three pack-scoped evidence subjects (logos_entry/gloss/morphology) with
logos_*:<packId>/<subId> address grammar, RightInspector projections, and
EvidenceChainRail derivations
- contents fetched eagerly on pack-select; /alignment stays deferred to LG-4
Full workbench-ui vitest: 506 passed, clean exit; build clean.
Read-only /logos frontend route (Wave M CORE-Logos read-only wave, LG-2):
- route registered via WORKBENCH_ROUTES + lazy ROUTE_ELEMENTS (Substrate section)
- Overview / Identity / Safety tabs over live /logos/packs, /{id}, /safety only
(no /contents or /alignment — those are LG-3 / LG-4)
- 10 Logos* TS interfaces mirrored; NOT_YET_MIRRORED shrunk back to empty
- SafetyVerdict UI badge + dump-enums.py source + enum-snapshot + coverage test
- logos_pack evidence subject (logos:<pack_id>) + round-trip
- bottom strip "proposal mode: none — read-only"; holonomy rendered as
missing_evidence (no tab, no proof card); invalid alignment targets surfaced
honestly in Safety; verdict never mapped to "clear" for warning/unknown
Depends on LG-1 (#737). No mutation endpoints, no engine math in the UI.
The workbench READMEs were frozen in the W-026..W-031 planning era. Bring them
current after the Wave 1 / R / M arc:
- docs/workbench/README.md (index): add the 5 unindexed docs (wave-1-evidence-
spine, wave-R-mastery-revamp, wave-M-worthiness, b4-leeway-producer-scope,
design-system); replace the false "Current Status" (which still pointed at the
superseded feat/w026 prototype) with the shipped state + a 14-route surface
table; mark the W-026..W-031 queue delivered; note Wave M complete (B+C+D+B4),
Phase E + parallel tracks remaining.
- workbench-ui/README.md: fix the stale one-line surface list and the "src/routes
placeholder components" description (routes are registry-driven in src/app/;
routes.ts is the single source). Left the dump-api-schemas.py reference intact
(it exists, distinct from dump-schemas.py).
- README.md (top level): the workbench section said "eleven routes" — now
fourteen, naming Tour, Contemplation, and Calibration and the per-turn
pipeline/field/leeway/bundle evidence.
Docs-only; no code or gate touched. UI-UX-GUIDE (the routes.docs.test-gated
route table) is unchanged and already at 14.
Clears the long-standing B4 block: the leeway decision was already made on the
serving path (chat/runtime.py::_surface_estimate has a real LicenseDecision +
the ReachPolicy) and then DISCARDED — never threaded to the result, and the
workbench can't import reliability_gate. This wires it through.
Producer (observational, never authorizing):
- core/cognition/leeway.py: LeewayRecord + build_leeway_record(reach_level,
license_decision) — duck-typed on the decision, zero new cross-package
coupling. Maps to: no decision -> "unknown" (STRICT, no latitude); denied
-> "blocked" (gate consulted, said no); licensed SERVE/PROPOSE widening ->
the real class / theta / "[approximate]" disclosure. "verified" is never
emitted (RESERVED state). source_digest content-addresses the decision
(deterministic, no wall-clock).
- core/cognition/result.py: additive `leeway: LeewayRecord | None = None` on
CognitiveTurnResult.
- core/cognition/pipeline.py: build it at result construction from the data
the runtime ALREADY exposes (response.reach_level +
runtime.last_turn_accrual().license). chat/runtime.py is UNTOUCHED.
- workbench/api.py: _leeway_evidence_from_result maps result.leeway ->
LeewayEvidence (pure projection; no reliability_gate import — firewall
intact). The journal already persists it; the B4a UI (Replay / Proposals /
RightInspector) already renders it — no frontend or schema change needed.
Safety (this touches the serving path, so proven, not asserted):
- trace_hash is a NAMED-field hash (core/cognition/trace.py); `leeway` is not
in it -> byte-identical serving. All provenance/trace tests pass.
- response_governance STRICT stays byte-identical (375 governance/serving/
provenance tests green, incl. the live-wiring + estimation-lane + ADR-0206
seam tests).
- core eval cognition: 13 cases, 100% intent / groundedness / versor closure.
- replay determinism holds (leeway is in CRITICAL_FIELDS; deterministic).
Tests: engine (build_leeway_record: strict/blocked/SERVE/PROPOSE, no "verified",
deterministic digest) + workbench mapping (field-for-field, honest absence,
invalid-enum clamp) + integration (a real turn now carries an honest leeway
record, not the null "No evidence recorded"). 151 workbench/leeway Python + 68
frontend (leeway/replay/proposals/schemaDrift) green; schema-snapshot unchanged.
Docs: gate cleared (b4-leeway-feasibility-gate.md), residue ledger flipped to
implemented, scope brief is b4-leeway-producer-scope-2026-06-13.md.
The leeway decision already exists at chat/runtime.py::_surface_estimate
(accrual.license is a real LicenseDecision) but is discarded — never threaded to
CognitiveTurnResult, and the workbench can't import reliability_gate. Brief lays
out: the exact producer seam, the LicenseDecision/ReachPolicy -> LeewayEvidence
mapping, two honest layers (STRICT-now / earned-APPROXIMATE), the firewall-safe
workbench mapping, constraints, a minimal additive Layer-1 first PR, and 4 open
questions.
The "they'd want to use it" first-run surface: a curated narrative that makes
the engine's discipline legible to a newcomer — bring a claim from any model and
watch it get decided, refused, or replayed. Proposer authority ignored.
Honest by construction (the risk with a narrative surface is theater):
- workbench/tour.py determinism_tour() is a curated ordered narrative BOUND to
the real demo registry. Intro (provider-agnostic thesis) → three demo steps
(deductive entailment decides only on engine+oracle agreement; epistemic
truth-state refuses a wrong proposer; proof-carrying promotion ignores
proposer authority) → payoff (replay-to-same-hash + the citable evidence
bundle).
- Each demo step's what_this_proves / what_this_does_not_prove cards and the
demo title are PULLED FROM THE REAL DEMO SPEC — never re-authored — so the
tour can never claim more than the demo it points at.
- A step referencing a missing demo FAILS CLOSED (KeyError), not a dead link.
Tests assert both (cards == spec; phantom demo raises).
Backend: schemas.py DeterminismTour/TourStep; GET /tour. schema-snapshot regen.
Frontend: /tour route registered in the registry (Determinism section, 14
routes; nav/palette/guide counts updated); TourRoute — thesis hero + ordered
step cards with the honesty cards + links to the real demos / replay.
Validation: 140 workbench Python tests incl. tour drift guards; 472/472 frontend
incl. routes/routeConformance/routes.docs (guide↔registry), Shell+palette counts
(14), schemaDrift, and the tour UI test; pnpm build clean; git diff --check
clean. No serving-path imports.
The "they'd want to use it" deliverable: a turn's evidence exported as ONE
deterministic, content-addressed, citable artifact — composing the Phase-C
evidence (pipeline + field) with the trace and the calibration leeway verdict.
Backend (read-only, no engine execution):
- schemas.py: EvidenceBundle.
- workbench/evidence_bundle.py: build_evidence_bundle(entry) assembles a turn
journal entry into the bundle and computes bundle_digest. The digest
content-addresses the DETERMINISTIC cognitive evidence only — journal
position + wall-clock (turn_id, journal_digest, replay_reproducer,
generated_from) are carried for provenance but EXCLUDED, so the same turn
content reproduces the same digest regardless of journal position.
- api.py: GET /trace/{turn_id}/bundle. schema-snapshot.json regenerated.
The bundle carries a replay *reproducer* command (how to verify) rather than a
live-run replay, so the artifact itself stays deterministic — verification is
the consumer's step: re-run the prompt sealed, confirm trace_hash, recompute the
bundle, check the digest.
Frontend:
- types/api.ts EvidenceBundle; client/query hook; Trace route **Bundle** tab —
citable digest, "what this proves / does not prove" honesty note, the
reproducer, and a deterministic JSON download (Blob anchor, leak-safe).
Validation: 135 workbench Python tests incl. non-vacuous bundle guards (digest
reproducible, journal-position/wall-clock excluded, any evidence change flips
the digest, missing Phase-C evidence honest); 468/468 frontend incl. schemaDrift
+ the citable-download bundle test; pnpm build clean; git diff --check clean. No
serving-path imports.
C2-a surfaced contemplation runs but rendered each scene as a flat JSON dump —
outputs, not process. C2-b makes the loop legible: every scene is now a typed
stage in the canonical ADR-0172 arc.
Reader-first (no theater):
- schemas.py: ContemplationScene gains a typed loop projection — stage_role
(cold_attempt | engine_enrichment | engine_proposal | operator_ratifies |
grounded | other) plus the connective ids (proposal_id, candidate_id,
proposal_state, grounding_source). New ContemplationStageRole Literal.
- readers.py: _contemplation_scenes derives stage_role from the scene id
(closed set, "other" fallback) and pulls the connective ids out of the raw
detail. schema-snapshot.json regenerated.
UI:
- ContemplationRoute: the run detail now renders the arc as a staged process —
"attempt → enrich → propose → ratify → grounded" with named stage roles,
cold→grounded bookends (surfaces pulled out of before/after), the connective
ids as evidence, and the raw detail tucked into a collapsible (not lost).
Honest wrinkle (surfaced, not faked): the fixture proposals do NOT resolve in
the live proposal log (source_kinds exemplar_corpus/operator, none
contemplation), so the proposal id is shown as evidence but is intentionally
NOT a clickable cross-route link — a dead link would be theater. Live
Proposals/Calibration navigation is deferred to reader-verified linking once
real contemplation proposals reach the log.
Validation: 128 workbench Python tests + new reader-projection tests (canonical
arc → roles, unknown → "other", id extraction, detail preserved); 466/466
frontend incl. schemaDrift + the staged-process / no-dead-link test; pnpm build
clean; git diff --check clean. No serving-path imports.
Makes the CL(4,1) field geometry legible, honestly. The engine owns the field;
this surfaces ONLY the exact scalar invariants it computes — never the raw
multivector — so the workbench shows "this is the geometry, it's exact, it
can't fake coherence" without a decorative blob or any motion.
Persist-first (the C3 gating work was Python, not React): the honest scalars
were computed live per turn but discarded. Now captured per turn into the
journal so a read-only surface has real evidence.
Backend:
- workbench/field_evidence.py: FieldEvidence computed from the engine result —
exact versor_condition, field_valid (vs the 1e-6 ceiling), a content-
addressed field_digest (sha256 of the engine-canonical array bytes), and
cga_inner(before, after) as the exact transition value. Raw field bytes
never cross the boundary: only floats + digests. validate() is fail-closed —
field_valid can never disagree with versor_condition vs the ceiling (the
wrong=0 analogue for the geometry). No engine math re-implemented
(versor_condition / cga_inner imported from algebra; bytes via array_codec).
- workbench/schemas.py: FieldEvidence dataclass; field_evidence on
ChatTurnResult + TurnJournalEntrySchema. schema-snapshot.json regenerated.
- workbench/journal.py + api.py: persisted at from_chat_turn; first-class read
endpoint GET /trace/{turn_id}/field (trace facet, consistent with /pipeline).
- workbench/replay.py: field_evidence classified CRITICAL — replay now also
proves field determinism (digest + scalars must match on re-execution).
Frontend:
- types/api.ts FieldEvidence + field_evidence passthrough; client/query hook;
FieldInvariantCard (measured value vs ceiling, cga_inner transition, digests;
honest missing_evidence; no blob, no motion); Trace route Field tab.
Honest-empty for pre-widening journal rows (missing_evidence). Deferred:
cross-turn field-coherence trends, session-level field persistence.
Validation: 138 workbench/practice Python tests (incl. non-vacuous field guards
+ replay field-determinism); 465/465 frontend incl. schemaDrift; pnpm build
clean; git diff --check clean. No generate.derivation / reliability_gate /
stream / field.propagate / vault.store imports.
Lands the Phase C "make cognition legible" slice plus Phase A residue, all
backend-reader-first over real engine data (no theater, read-only doctrine
intact, zero serving-path imports).
C1-a — Cognitive pipeline record (persistence-first, per #729 worthiness edit):
- workbench/pipeline_record.py: curated CognitivePipelineRecord over the real
CognitiveTurnResult (input → intent → proposition_graph → articulation_target
→ realizer → walk_telemetry → trace_hash). Raw field multivectors are
DELIBERATELY excluded; _assert_no_raw_field_payload recursively rejects raw
field keys, and validate_pipeline_record fails closed on missing/duplicate
stages, non-recorded status, or dangling edges — the UI can never receive a
partial record that claims to be complete.
- test_workbench_pipeline_record.py: non-vacuous guards — missing stage,
monkeypatched new required stage, and injected raw {"F": [...]} each raise.
C2-a — Contemplation as a process: /contemplation route over real persisted
contemplation/runs/*.json (glob reader; honest-empty when absent).
C4-a — Identity continuity (L10/L11): RunDetail.identity_continuity + Runs
Identity tab, sourced from the real core.engine_identity (engine_identity /
parent_engine_identity lineage relation, re-derived to verify).
Demo Theater: renders backend-owned proof-promotion + entailment DAGs.
Phase A residue: density preference wired end-to-end (settings → shell → tokens);
cross-route consistency touch-ups.
Infra: local API CORS now echoes only validated 127.0.0.1/localhost origins
(hostname-checked, not arbitrary reflection) so Vite fallback ports work.
Route chunk-split keeps the build warning-free.
Cleanup: corrected the stale ADR-0175 practice-lane assertions (build_report is
6 correct / 0 wrong / 44 refused after the current serving lane; wrong=0 held)
and the two registry-derived count tests (LeftNav + CommandPalette 12 → 13 for
the new Contemplation route).
Docs: runtime_contracts.md (pipeline-record contract), UI-UX-GUIDE,
api-contract-v1, data-shapes-v1, wave-M-worthiness, phase-a-residue-ledger.
Validation: 106 workbench/practice Python tests green (incl. wrong=0 lane +
pipeline-record fail-closed guards); 459/459 frontend; pnpm build clean;
git diff --check clean. No generate.derivation / reliability_gate / stream /
field.propagate / vault.store imports.
Completes the Wave M B3.5 consolidation slice (b–e), built on #728.
B3.5-b — calibration as a first-class evidence subject (`calibration_class`,
address `calibration:<class_name>`): RightInspector projection + Evidence
Chain Rail semantics (serving-discipline evidence, not runtime truth).
B3.5-c / B4a — nullable `LeewayEvidence` read model threaded through turn,
replay, cognition-proposal, and math-proposal surfaces, with a shared
absence-honest card. B4 is gated correctly: the tuple exists in typed data but
no producer populates it, so the card renders absence (verified: no non-null
producer in workbench/core/chat).
B3.5-d/e — UI-UX-GUIDE.md, b4-leeway-feasibility-gate.md, phase-a-residue-ledger.md.
Practice artifact — earn-it-for-real (runner-reproducible). The committed
`report.json` (additive earns PROPOSE @0.861, 95/5/50) is now emitted by a
deterministic runner rather than copied from the queue. `propose_runner`
gains `regenerate_practice_artifacts()`, which runs ONE sealed `resolve_pooled`
practice pass and writes BOTH report.json (the per-class ledger the calibration
reader consumes) and ratification_queue.json — two projections of one ledger,
coherent by construction and byte-reproducible. `runner.main()` delegates to
it (lazy import, no cycle), so both entry points produce the identical pair.
This closes the gap where a hand-copied report.json agreed with the queue but
no runner produced it. `resolve_pooled` is the aggressive sealed PROPOSE-regime
scorer (proposal-only/HITL, unsafe for serving, legitimate for
attempt-and-eliminate); wrong=5 is the sealed-practice learning signal, NOT the
serving wrong=0. No serving/derivation/reliability_gate source touched; the
practice lane is not in the serving-frozen SHA gate.
Validated:
- python -m pytest tests/test_workbench_{calibration,journal,replay,schemas}.py -> 31 passed
- python -m pytest tests/ -k "workbench or propose or learning_arena or practice"
-> 190 passed (3 failing tests in test_adr_0175_phase2_practice_lane.py are
PRE-EXISTING reds on clean origin/main: stale 4/0/46 assertions on build_report,
which this change does not touch)
- report.json + ratification_queue.json: deterministic (run1==run2) and
reproduced byte-identically by both `python -m ...runner` and `...propose_runner`
- pnpm build green; 144 UI tests across calibration/leeway/evidence/replay/
doctrine-gates/routes-docs-drift all pass
Two corrections from this session's source verification:
- B3.5 D2 (calibration artifact): coherence between report.json and
ratification_queue.json is necessary but NOT sufficient. The 95/50/5 earned
state is a fossil from the disabled resolve_pooled scorer (queue committed in
b82897a0); no current runner reproduces it (main()=6/44/0, candidate-graph
practice=0/1/149). Acceptance now requires report.json be byte-reproducible
by a documented deterministic runner pass; copying numbers between artifacts
is inadmissible. Three routes named (honest-floor now, earn-it-for-real
follow-up, defer).
- C1 (cognitive pipeline visualizer): corrected the false "reader-first over
existing telemetry" framing. Verified the journal persists only ~12 surface
fields of CognitiveTurnResult's ~25; no /trace/{id}/pipeline endpoint. C1-a's
real first deliverable is a persistence change: a curated
CognitivePipelineRecord (never the raw field_state multivectors — O(n^2)
cost), a runtime-contract update, and a non-vacuous fail-closed test.
Adds `workbench-ui/src/app/routes.ts` as the single source of truth for
navigation routes. App, LeftNav, the command palette, ⌘-digit shortcuts,
and the landing-route preference now all derive from `WORKBENCH_ROUTES`.
Fixes the live palette-drift bug: the command palette's hand-maintained
NAV_COMMANDS listed 10 routes and silently dropped Demos and Calibration,
so they were unreachable via ⌘K. The landing-route tuple was likewise
missing Replay and Calibration. Deriving every list from one registry
makes that drift structurally impossible.
- routes.ts carries a `section` field for grouped wayfinding nav
(Converse · Cognition · Determinism · Evidence · Discipline ·
Substrate · Settings) — a display skin only; one workbench, one
address space, one Evidence Chain Rail.
- App is data-driven via ROUTE_ELEMENTS (route id → element); a registry
route without an element fails routes.test rather than rendering undefined.
- Honest keyboard model: ten routes pin ⌘1–0; Demos and Calibration are
palette-only (no chord advertised).
- routes.test.tsx guards the contract: element-map parity, unique
ids/paths/digits, landing-route coverage (Replay + Calibration), and
palette reachability of every command-visible route.
Verified: pnpm build green; routes/Shell/CommandPalette/routeConformance/
prefs/Settings + hexScan/schemaDrift/enumCoverage suites all pass (108 tests).
The docs/proposal-artifact-substrate-v1 work (#727) — wave-m-consolidation-b3.5,
core-logos-studio-plan, proposal-artifact-substrate-v1 — is the authoritative
consolidation and supersedes the parallel 'Phase A' I'd just drafted. Reconciling
to one source of truth:
- FOLD IN the one genuine gap b3.5 missed (a design review caught it): the
Calibration centerpiece undersells — the reader reads the sub-N_MIN committed
practice report.json, so on live data no class shows licensed, while the
earned state (additive 95/5/100, PROPOSE-licensed) sits in the disagreeing
ratification_queue.json. Added as a diagnosis addendum + a Deliverable-2
acceptance criterion (make the committed practice artifacts coherent; show
≥1 earned class). Data-side fix, not a re-derivation.
- FOLD IN the grouped-navigation idea: a field on the route registry
(Deliverable 1) + the organism's-loop grouping (Converse/Cognition/Evidence/
Determinism/Discipline/Substrate/Settings), with the correction that
'core-logos' is the language/manifold Studio surface, not the cognition cluster.
- REDIRECT wave-M-worthiness.md's consolidation section to point at b3.5 instead
of defining a parallel Phase A.
- SCRAP duplicates: wave-M-phaseA-briefs (duplicated b3.5) and
workbench-ui-ux-guide-brief (superseded by b3.5 Deliverable 4).
Boils the design review and the information-architecture decision into one
integration. Phase B's heart (calibration/serving discipline) is done;
Phase A (Structure & Polish) now runs BEFORE resuming Phase C because two of
its items are structural prerequisites the Cognition cluster lands into.
Key synthesis: the command-palette drift bug and the grouped-nav idea are
the SAME fix — navigation derives from one registry that also encodes
structure. Standing IA constraint: one workbench, one address space, one
Chain Rail; grouping is a wayfinding skin, never an architectural fork; no
separate workbenches (the single evidence model is the thesis); 'levels' are
depth-within-a-surface, not top-level categories.
Phase A brief pack (wave-M-phaseA-briefs):
- A1: navigation registry (one source of truth) + grouped nav by the
organism's loop (Converse/Cognition=core-logos/Evidence/Determinism/
Discipline/Substrate/Settings) + palette fix (Demos+Calibration are
currently unreachable via ⌘K; the ⌘1-0 map is stale)
- A2: Calibration earned-state — the centerpiece undersells (reads the
sub-N_MIN report.json; the earned license lives in the disagreeing
ratification_queue). Primary fix: regenerate coherent practice artifacts;
fallback: reader surfaces both with provenance
- A3: Doctrine station ('how this UI can't lie') — elevate contracts/checks
into a first-class surface; trails, larger
Parked: B4 (needs engine-side license stamping; no re-derivation),
Calibration EvidenceSubject kind.
An always-present invariant in the TopBar chrome: the live serving triplet
(N correct · N refused · 0 wrong) aggregated across lanes from
/serving/metrics, the wrong count load-bearing — 0 in the verified token,
>0 in the contradicted token. It is a MIRROR, never a hard-coded zero:
renders a non-zero wrong honestly, and an honest 'metrics unavailable' when
the reports can't be read. Links to the Calibration route (the discipline
behind the number).
Tests: aggregate triplet + 0-wrong, non-zero-wrong-honest (not hard-coded),
metrics-unavailable-not-fake-zero. (Adds fetchServingMetrics/useServingMetrics
locally; unioned with B2 on rebase.)
Consumes the B1 readers (#724). The route where you SEE 'the engine earns
the right to guess' (ADR-0175).
- list (failures-first, server-ordered): per class a reliability bar
(Wilson floor vs the PROPOSE/SERVE θ markers — a class earns the right
when its fill crosses the marker), correct/refused/wrong counts (wrong in
the contradicted token), and a verdict pill (earned SERVE / earned
PROPOSE / not yet licensed)
- serving strip: the discipline's RESULT — live correct/refused/wrong per
lane, wrong=0 in the verified token, with source DigestBadges
- detail (Panel + TabBar): Counts / License math / Raw. The License-math
tab shows the honest derivation (measured ≥ θ → licensed) and states
the numbers come from core.reliability_gate, not the workbench
- fail-closed: an empty/absent arena ledger (501 evidence_unavailable)
renders the honest absence card pointing at the practice lane, not an
error
- nav entry (12th) + routeConformance row + Shell nav assertion updated
Token-only (hexScan green); VirtualizedList + j/k; full vitest 398 green.
First Wave M / Phase B piece (GATING): read-only backend that makes the
calibrated-learning / serving-discipline loop inspectable — 'the engine
earns the right to guess', ADR-0175.
The workbench computes NONE of these numbers:
- GET /calibration/classes — per-class gold-tether view from the persisted
practice arena ledger (evals/gsm8k_math/practice/v1/report.json per_class).
Each class's reliability_floor is the engine's own one-sided Wilson
conservative_floor (via ClassTally.reliability); PROPOSE (θ=0.85) / SERVE
(θ=0.99) license verdicts come from core.reliability_gate.license_for.
Failures-first ordering. A test proves the reader's floor equals a direct
conservative_floor() call — no re-implementation.
- GET /serving/metrics — the live correct/refused/wrong counts read unchanged
from the committed train_sample + holdout_dev report.json (currently
4/46/0 and 5/495/0 — wrong=0). Never re-runs a lane.
Honest current state: the committed practice ledger's three classes
(additive/divisive/multiplicative) are all below N_MIN=10, so none has
earned a license yet — the reader shows exactly that, no fake green light.
- workbench/calibration.py: pure readers; imports core.reliability_gate;
EvidenceUnavailableError -> 501 (fail-closed) when the artifact is absent.
- schemas + TS mirrors (CalibrationClass, ServingMetrics); both snapshots
regenerated (deterministic); both drift gates pass.
- trust boundary: read-only over committed artifacts + engine-owned
derivation; no execution, no mutation, no license ever changed.
Verified: 30 Python tests (incl. the no-reimplementation proof + fail-closed),
390 vitest, both schema drift gates, snapshots deterministic.
